diff --git a/.agents/notes/implemented/architecture/2026-07-28-dsh-native-typescript-source-launch.i18n.yaml b/.agents/notes/implemented/architecture/2026-07-28-dsh-native-typescript-source-launch.i18n.yaml new file mode 100644 index 0000000000..a96c124e92 --- /dev/null +++ b/.agents/notes/implemented/architecture/2026-07-28-dsh-native-typescript-source-launch.i18n.yaml @@ -0,0 +1,6 @@ +# Bilingual-pair consistency record (docs/i18n/README.md): the git blob hash of each +# side as of the last confirmed-consistent state. Both languages carry equal authority; +# after editing either side, bring the other along and re-record with: +# pnpm run verify-translation-pairing --write .agents/notes/implemented/architecture/2026-07-28-dsh-native-typescript-source-launch.md +2026-07-28-dsh-native-typescript-source-launch.md: 019389f3e5e9229f4359bbd58c95dbb2f14eb24b +2026-07-28-dsh-native-typescript-source-launch.zh.md: 2cfff25d228e67ac85a9bc9087fa09ddb64213a0 diff --git a/.agents/notes/implemented/architecture/2026-07-28-dsh-native-typescript-source-launch.md b/.agents/notes/implemented/architecture/2026-07-28-dsh-native-typescript-source-launch.md new file mode 100644 index 0000000000..019389f3e5 --- /dev/null +++ b/.agents/notes/implemented/architecture/2026-07-28-dsh-native-typescript-source-launch.md @@ -0,0 +1,45 @@ +# Agent Note: Native TypeScript source launch for dsh + +Status: implemented + +English | [中文](2026-07-28-dsh-native-typescript-source-launch.zh.md) + +## Problem + +The `dsh` source entry point originally used `tsx` to run `apps/cli/src/bin.ts`, with the same third-party loader implicitly handling both TypeScript transformation and the root tsconfig's `paths` resolution. With Node handling TypeScript natively, it does not apply tsconfig path mappings; resolving through package exports would instead mix potentially stale or nonexistent `lib/` artifacts into the source launch. + +Node's transform also does not perform type analysis. A type imported through an ordinary value import remains a runtime ESM request, and TypeScript's `export =` becomes a CommonJS assignment rather than an ESM default export. The source graph therefore has to use explicit type-only imports and native ESM exports; a resolve hook cannot repair incompatible source syntax. + +Cordis configuration introduces a separate resolution boundary. Bare plugins in `cordis.yml` do not pass through TypeScript import analysis, so their resolver manifest may omit the required dependencies. The Cordis Loader logs plugin import errors and leaves an entry without a fiber, but does not fail startup itself; a typo in the configuration can therefore produce an incomplete application with exit code 0. + +## Decision + +The `dsh` TUI, Web, and headless source launches use `node --experimental-transform-types`; Node performs TypeScript transformation without loading `tsx` or esbuild. `bin/dsh`, the root-level `dsh`/TUI/Web demos, and Code Mode TUI enter the same `apps/cli/src/bin.ts` launch chain. Test and E2E launchers retain their existing strategies, and the built `lib/bin.js` continues to run under ordinary Node. + +`scripts/tspath-loader.ts` registers only a module resolve hook. It uses `TSX_TSCONFIG_PATH` when set (resolving relative values from the invoking cwd) and otherwise reads the root `tsconfig.json`; `TsconfigPathsResolver` follows that config's `extends` chain through the repository's existing TypeScript development tool, selects exact or wildcard `paths` entries according to tsconfig rules, and maps matching workspace bare specifiers to `.ts`/`.mts`/`.cts` source files or directory index files. Node remains solely responsible for code transformation. The source-only loader is not part of the built CLI and `apps/cli` does not declare `typescript` as a runtime dependency. + +Source imports are redirected only when the target package is either the nearest package manifest's own name or one of that manifest's declared runtime dependencies. The Cordis Loader uses the configuration directory URL as the import parent; the resolver then searches upward for the workspace manifest that declares the plugin, so dependency ownership for `examples/tui-agent/cordis.yml` lies with `examples/package.json`, and dependency ownership for `apps/cli/cordis.yml` lies with `apps/cli/package.json`. Specifiers that do not match tsconfig paths, refer to undeclared dependencies, or are not bare all fall back to Node's default resolution. + +`verify-cordis-config` performs a one-way completeness check on both resolver manifests: every bare plugin package in a configuration must appear in the corresponding manifest's `dependencies`, while the manifest may contain extra dependencies not referenced by that configuration. The root `AGENTS.md` makes updating the configuration and dependencies together a standing rule. + +After the Loader settles, the shared `dsh-app-boot` checks every enabled entry that has no fiber and rejects startup with `plugin(s) failed to load: ...; Cordis startup failed because these plugin(s) could not be resolved`, listing all failed plugins. This diagnostic lives at the app layer and does not change the vendored Loader's startup behavior. + +Node-compatible TypeScript is part of this source-launch contract. Vendored Cordis, Loader, Include, HMR, and Schemastery mark erased imports with `import type`. Schemastery uses a native ESM default export and declares `type: module`; its `.mjs` and `.cjs` build outputs retain the existing ESM-default and callable-`require()` behavior. These divergences are recorded in `vendor/README.md`; no runtime behavior is added to the vendored frameworks. + +## Alternatives considered + +**Continue using `tsx`.** Rejected because `tsx`/esbuild would continue to own TypeScript transformation, so this launch chain could not prove that Node's native transformation works. + +**Load the built `lib/` through package exports from the source entry point.** Rejected because this would mix the source plane with the artifact plane; a zero-build development launch could read stale artifacts or fail outright. + +**Apply the root tsconfig `paths` unconditionally.** Rejected because this would allow undeclared cross-package imports and Cordis plugins to keep resolving, hiding mismatches between the manifest and the actual runtime graph. + +**Transform imports inside the custom loader.** Rejected because type-aware source rewriting would reintroduce a compiler-style transform and make the loader, rather than Node, responsible for TypeScript execution. Making the checked-in source Node-compatible keeps the launch boundary explicit. + +## Consequences + +- TUI/headless retain a zero-build source loop, while Web still builds its frontend artifacts before starting the CLI source entry. TypeScript syntax passes only through Node's native transform; the URL-only loader uses the checkout's root development dependencies and adds no CLI runtime dependency. +- Workspace package imports and Cordis configuration dependencies must both be declared explicitly in the resolver manifest; the static gate prevents configuration from landing before its dependencies, while extra dependencies are not errors. +- Plugin import failures no longer leave an incomplete application with exit code 0; the final error identifies both the Cordis startup failure and the specific plugin names, while the Loader's original error remains earlier in the logs. +- Vendored source in the CLI graph must remain compatible with Node's transform-types module semantics; the local-modification log makes the upstream sync obligation explicit. +- CI's `lib` mode, test/E2E launchers, and other example launchers retain their existing strategies; this native source loader covers only the `dsh` CLI application chain. diff --git a/.agents/notes/implemented/architecture/2026-07-28-dsh-native-typescript-source-launch.zh.md b/.agents/notes/implemented/architecture/2026-07-28-dsh-native-typescript-source-launch.zh.md new file mode 100644 index 0000000000..2cfff25d22 --- /dev/null +++ b/.agents/notes/implemented/architecture/2026-07-28-dsh-native-typescript-source-launch.zh.md @@ -0,0 +1,45 @@ +# Agent Note: dsh 原生 TypeScript 源码启动 + +Status: implemented + +[English](2026-07-28-dsh-native-typescript-source-launch.md) | 中文 + +## 问题 + +`dsh` 源码入口原本使用 `tsx` 运行 `apps/cli/src/bin.ts`，TypeScript 转换和根 tsconfig 的 `paths` 解析都由同一个第三方 loader 隐式处理。改由 Node 原生处理 TypeScript 后，Node 不会应用 tsconfig 路径映射；如果改为通过包导出解析，源码启动会混入可能陈旧或不存在的 `lib/` 产物。 + +Node 的转换也不执行类型分析。通过普通值 import 导入的类型会保留为运行时 ESM 请求，而 TypeScript 的 `export =` 会变成 CommonJS 赋值，而不是 ESM default export。因此，源码图必须显式使用仅类型导入和原生 ESM 导出；resolve hook 无法修复不兼容的源码语法。 + +Cordis 配置还引入了另一条解析边界。`cordis.yml` 中的 bare plugin 不经过 TypeScript import 分析，其解析方 manifest 可能漏掉所需依赖。Cordis Loader 会记录插件 import 错误，并留下没有 fiber 的 entry，但不会让启动本身失败；配置中的拼写错误因此可能得到退出码为 0 的残缺应用。 + +## 决策 + +`dsh` 的 TUI、Web 和无头源码启动使用 `node --experimental-transform-types`，由 Node 完成 TypeScript 转换，不加载 `tsx` 或 esbuild。`bin/dsh`、根级 `dsh`/TUI/Web demo 以及 Code Mode TUI 都进入同一条 `apps/cli/src/bin.ts` 启动链路。测试与 e2e 启动器保留各自现有策略，构建后的 `lib/bin.js` 继续由普通 Node 运行。 + +`scripts/tspath-loader.ts` 只注册一个模块 resolve hook。设置 `TSX_TSCONFIG_PATH` 时，它会使用该路径（相对路径从调用方的 cwd 解析），否则读取根 `tsconfig.json`；`TsconfigPathsResolver` 使用仓库已有的 TypeScript 开发工具沿该配置的 `extends` 链解析，按 tsconfig 规则选择精确或 wildcard `paths` 条目，并将命中的 workspace bare specifier 映射到 `.ts`/`.mts`/`.cts` 源文件或目录 index 文件。代码转换始终只由 Node 负责。该源码专用 loader 不属于构建后的 CLI，`apps/cli` 也不会把 `typescript` 声明为运行时依赖。 + +只有当目标包是最近 package manifest 的自身名称或其已声明的运行时依赖时，源码 import 才会重定向。Cordis Loader 使用配置目录 URL 作为 import parent；此时 resolver 会向上查找声明该插件的 workspace manifest。因此，`examples/tui-agent/cordis.yml` 的依赖由 `examples/package.json` 持有，`apps/cli/cordis.yml` 的依赖由 `apps/cli/package.json` 持有。未命中 tsconfig paths、引用未声明依赖或不是 bare specifier 的说明符全部交回 Node 默认解析。 + +`verify-cordis-config` 对这两个解析方 manifest 执行单向完整性检查：配置中的每个 bare plugin package 都必须出现在对应 manifest 的 `dependencies` 中，manifest 可以包含该配置未引用的额外依赖。根 `AGENTS.md` 将同步更新配置和依赖定为常驻规则。 + +Loader 完成结算后，共享的 `dsh-app-boot` 会检查每个已启用但没有 fiber 的 entry，并以 `plugin(s) failed to load: ...; Cordis startup failed because these plugin(s) could not be resolved` 拒绝启动，同时列出全部加载失败的插件。该诊断位于应用层，不改变 vendor 中 Loader 的启动行为。 + +Node-compatible TypeScript 是这项源码启动契约的一部分。vendor 中的 Cordis、Loader、Include、HMR（热模块替换）和 Schemastery 使用 `import type` 标记会被擦除的导入。Schemastery 使用原生 ESM default export 并声明 `type: module`；其 `.mjs` 和 `.cjs` 构建产物分别保留现有的 ESM default export 行为和 `require()` 返回可调用值的行为。这些差异记录在 `vendor/README.md` 中；没有为 vendor 中的框架新增运行时行为。 + +## 曾考虑的替代方案 + +**继续使用 `tsx`。** 不采用，因为 `tsx`/esbuild 会继续负责 TypeScript 转换，本启动链路无法因此证明 Node 原生转换可用。 + +**让源码入口通过包导出加载构建后的 `lib/`。** 不采用，因为这会混合 source plane 与 artifact plane；零构建开发启动可能读取陈旧产物或直接失败。 + +**无条件应用根 tsconfig `paths`。** 不采用，因为这会让未声明的跨包 import 和 Cordis 插件继续成功解析，从而掩盖 manifest 与实际运行图之间的不一致。 + +**在自定义 loader 内转换 import。** 不采用，因为感知类型的源码改写会重新引入编译器式转换，并让 loader 而非 Node 负责执行 TypeScript。使签入仓库的源码兼容 Node，可以让启动边界保持显式。 + +## 后果 + +- TUI／无头界面保留零构建源码回路，Web 仍会在启动 CLI 源码入口前构建前端产物。TypeScript 语法只经过 Node 原生转换；仅处理 URL 的 loader 使用 checkout 根目录的开发依赖，不增加 CLI 运行时依赖。 +- workspace package import 和 Cordis 配置依赖都必须在解析方 manifest 中明确声明；静态门禁防止配置先于依赖落地，额外依赖不构成错误。 +- 插件 import 失败不再留下退出码为 0 的残缺应用；最终错误同时说明 Cordis 启动失败及具体插件名，Loader 的原始错误仍会保留在更早的日志中。 +- CLI 源码图中的 vendor 源码必须与 Node 的 transform-types 模块语义兼容；本地修改记录明确了上游同步义务。 +- CI 的 `lib` 模式、测试／e2e 启动器和其他示例启动器保留各自现有策略；该原生源码 loader 只覆盖 `dsh` CLI 应用链路。 diff --git a/.agents/notes/implemented/bug-fix/2026-07-29-pnpm-setup-runner-isolation.i18n.yaml b/.agents/notes/implemented/bug-fix/2026-07-29-pnpm-setup-runner-isolation.i18n.yaml new file mode 100644 index 0000000000..fb5ee5debc --- /dev/null +++ b/.agents/notes/implemented/bug-fix/2026-07-29-pnpm-setup-runner-isolation.i18n.yaml @@ -0,0 +1,6 @@ +# Bilingual-pair consistency record (docs/i18n/README.md): the git blob hash of each +# side as of the last confirmed-consistent state. Both languages carry equal authority; +# after editing either side, bring the other along and re-record with: +# pnpm run verify-translation-pairing --write .agents/notes/implemented/bug-fix/2026-07-29-pnpm-setup-runner-isolation.md +2026-07-29-pnpm-setup-runner-isolation.md: 743535d0394cbea0374c412ba6968910ce858de4 +2026-07-29-pnpm-setup-runner-isolation.zh.md: 1e51070f88dead17b9d3f5625e337c558786aba2 diff --git a/.agents/notes/implemented/bug-fix/2026-07-29-pnpm-setup-runner-isolation.md b/.agents/notes/implemented/bug-fix/2026-07-29-pnpm-setup-runner-isolation.md new file mode 100644 index 0000000000..743535d039 --- /dev/null +++ b/.agents/notes/implemented/bug-fix/2026-07-29-pnpm-setup-runner-isolation.md @@ -0,0 +1,27 @@ +# Agent Note: Isolate pnpm setup per GitHub Actions runner + +Status: implemented + +English | [中文](2026-07-29-pnpm-setup-runner-isolation.zh.md) + +## Problem + +`pnpm/action-setup@v4` defaults its install destination to `~/setup-pnpm` and replaces that directory during setup. The self-hosted CI failover runs six GitHub Actions runner services under one VM user, so concurrent jobs shared the same destination. In [run 30375670773](https://github.com/deepseek-harness/deepseek-harness/actions/runs/30375670773), three jobs entered pnpm setup within 73 milliseconds; one setup removed another process's current working directory and two jobs failed in Node's `uv_cwd` initialization. A retry on another runner passed, making the failure timing-dependent rather than a repository-test regression. + +## Decision + +Every `pnpm/action-setup` step in [the primary CI workflow](../../../../.github/workflows/ci.yml) sets `dest: ${{ runner.temp }}/setup-pnpm`. Each runner service owns its temporary directory, so one setup cannot replace another runner's install directory. Persistent store reuse remains separate through `PNPM_CONFIG_STORE_DIR`, as established by the [pnpm provisioning decision](../process/2026-07-26-pnpm-action-setup-for-symmetric-ci-caching.md). + +[The workflow regression test](../../../../scripts/ci-workflow.spec.ts) discovers every `pnpm/action-setup` step in `ci.yml` and rejects one without the runner-private destination. This keeps newly added jobs inside the same isolation boundary. + +## Alternatives considered + +**Serialize failover jobs.** Rejected because it discards the six-runner pool's intended parallelism and turns an action-local directory collision into queueing across otherwise independent jobs. + +**Assign a separate Unix user to every runner service.** This would also separate `HOME`, but it moves the invariant into external VM provisioning and complicates ownership of the deliberately shared persistent pnpm store. The workflow already receives a runner-private temporary directory. + +**Retry failed setup steps.** Rejected because retries only reduce the observed collision rate; another concurrent setup can remove the same shared directory again. + +## Consequences + +pnpm's executable installation is ephemeral and isolated per runner, while package downloads still use the configured persistent or cached store. Hosted jobs use the same explicit destination without changing cache policy. The workflow carries three extra configuration lines per setup step, and the regression test must be updated only if pnpm provisioning intentionally moves to a different isolation mechanism. diff --git a/.agents/notes/implemented/bug-fix/2026-07-29-pnpm-setup-runner-isolation.zh.md b/.agents/notes/implemented/bug-fix/2026-07-29-pnpm-setup-runner-isolation.zh.md new file mode 100644 index 0000000000..1e51070f88 --- /dev/null +++ b/.agents/notes/implemented/bug-fix/2026-07-29-pnpm-setup-runner-isolation.zh.md @@ -0,0 +1,27 @@ +# Agent Note: 按 GitHub Actions runner 隔离 pnpm 设置 + +Status: implemented + +[English](2026-07-29-pnpm-setup-runner-isolation.md) | 中文 + +## 问题 + +`pnpm/action-setup@v4` 的安装目标目录默认为 `~/setup-pnpm`，并会在设置期间替换该目录。自托管 CI 故障切换在同一个 VM 用户下运行六个 GitHub Actions runner 服务，因此并发作业会共用同一目标目录。在 [run 30375670773](https://github.com/deepseek-harness/deepseek-harness/actions/runs/30375670773) 中，三个作业在 73 毫秒内进入 pnpm 设置；其中一个设置过程删除了另一个进程的当前工作目录，导致两个作业在 Node 的 `uv_cwd` 初始化阶段失败。换到另一台 runner 重试后通过，说明该故障取决于时序，并非仓库测试回归。 + +## 决策 + +[主 CI 工作流](../../../../.github/workflows/ci.yml)中的每个 `pnpm/action-setup` 步骤都设置 `dest: ${{ runner.temp }}/setup-pnpm`。每个 runner 服务独占自己的临时目录，因此一个设置过程无法替换另一个 runner 的安装目录。持久 store 的复用仍由 `PNPM_CONFIG_STORE_DIR` 独立处理，遵循 [pnpm 提供机制决策](../process/2026-07-26-pnpm-action-setup-for-symmetric-ci-caching.md)。 + +[工作流回归测试](../../../../scripts/ci-workflow.spec.ts)会找出 `ci.yml` 中的每个 `pnpm/action-setup` 步骤，并拒绝缺少 runner 专属目标目录的步骤。这可确保后续新增的作业也处于同一隔离边界内。 + +## 曾考虑的替代方案 + +**串行执行故障切换作业。** 否决：这会牺牲由六个 runner 组成的池所具备的预期并行能力，并把 action 内部的目录冲突变成原本相互独立作业之间的队列等待。 + +**为每个 runner 服务分配独立的 Unix 用户。** 这同样能够隔离 `HOME`，但会把该不变量转移到外部 VM 配置中，并使刻意共享的持久 pnpm store 的所有权变得复杂。工作流已经获得 runner 专属临时目录。 + +**重试失败的设置步骤。** 否决：重试只能降低观测到的冲突发生率；另一个并发设置过程仍可能再次删除同一个共享目录。 + +## 后果 + +pnpm 可执行文件采用临时安装，并按 runner 隔离；包下载仍使用已配置的持久或缓存 store。托管作业使用相同的显式目标目录，不改变缓存政策。工作流中的每个设置步骤因此增加三行配置；只有在 pnpm 提供机制有意迁移到另一种隔离机制时，才需要更新回归测试。 diff --git a/.agents/notes/implemented/feature/2026-07-06-approval-seam.i18n.yaml b/.agents/notes/implemented/feature/2026-07-06-approval-seam.i18n.yaml index d225ddf2d5..51ddb7e70f 100644 --- a/.agents/notes/implemented/feature/2026-07-06-approval-seam.i18n.yaml +++ b/.agents/notes/implemented/feature/2026-07-06-approval-seam.i18n.yaml @@ -1,6 +1,6 @@ # Bilingual-pair consistency record (docs/i18n/README.md): the git blob hash of each # side as of the last confirmed-consistent state. Both languages carry equal authority; # after editing either side, bring the other along and re-record with: -# pnpm run verify-translation-pairing --write -2026-07-06-approval-seam.md: 70ccd4d486ad6e0126fa2eb638a064e9fc89bba6 -2026-07-06-approval-seam.zh.md: d218f79888957735305db14cd97cc74480297d29 +# pnpm run verify-translation-pairing --write .agents/notes/implemented/feature/2026-07-06-approval-seam.md +2026-07-06-approval-seam.md: efb4159d736779af28edc1ae6091de4669c92f31 +2026-07-06-approval-seam.zh.md: 9a4656f30a43473fe90cde9c56d45f48943e5d10 diff --git a/.agents/notes/implemented/feature/2026-07-06-approval-seam.md b/.agents/notes/implemented/feature/2026-07-06-approval-seam.md index 70ccd4d486..efb4159d73 100644 --- a/.agents/notes/implemented/feature/2026-07-06-approval-seam.md +++ b/.agents/notes/implemented/feature/2026-07-06-approval-seam.md @@ -123,7 +123,7 @@ Costs and accepted limits: - **Who decides whether a call asks in the first place?** Policy producers: a hook returning `permissionDecision: ask`, any `tools/pre-execute` listener, or the sandbox escalation gate. The seam and the bridge only route and answer; neither injects its own judgment about what deserves a prompt. - **What happens when the user dismisses the prompt, or the turn aborts mid-ask?** Dismissal maps to `cancelled` with its own deny text. An already-aborted signal settles `cancelled` without dispatching; an abort during the ask discards the late answer. When both audit appends commit, either path records one pair, never two. - **What if the client answers with an option the harness never offered?** Any selection other than the offered `allow_once` maps to `rejected` — an unknown optionId from a non-conforming client can never grant. -- **How do subagents' approvals route?** An agent no answerer owns delegates through the whole waterfall and fails closed — in-process subagents are deliberately unanswerable. `subagent-acp`'s child-side auto-answer is separate; routing a child's asks to the parent controller is deferred (§ Deferred). +- **How do subagents' approvals route?** An agent no answerer owns delegates through the whole waterfall and fails closed — in-process subagents are deliberately unanswerable. A `'never'` parent seeds that override into each in-process child's log ([decision](2026-07-25-subagent-policy-inheritance.md)), so the child is told up front instead of asking into the empty waterfall. `subagent-acp`'s child-side auto-answer is separate; routing a child's asks to the parent controller is deferred (§ Deferred). - **What does `policy: 'never'` actually change at runtime?** The service resolves every ask for that session to `rejected` before dispatching any answerer (in-service, so no registration order can bypass it); the system prompt states the policy; switches are narrated at boundaries; each successful auto-rejection records the audit pair. - **What happens across a hot reload, or when an answerer unloads mid-session?** Answerers dispose with their owning fiber, so the next ask degrades to `unavailable` instead of hanging on a dead channel; remounting re-registers the answerer with no catch-up state. - **Where does a client get approval context?** The request carries the exact `callId` and the asker's human-readable `reason`; channel adapters may correlate richer tool-call state without duplicating arguments in the approval seam. diff --git a/.agents/notes/implemented/feature/2026-07-06-approval-seam.zh.md b/.agents/notes/implemented/feature/2026-07-06-approval-seam.zh.md index d218f79888..9a4656f30a 100644 --- a/.agents/notes/implemented/feature/2026-07-06-approval-seam.zh.md +++ b/.agents/notes/implemented/feature/2026-07-06-approval-seam.zh.md @@ -123,7 +123,7 @@ ACP 桥只应答其会话映射所拥有的精确 agent 对象。它携带既有 - **谁决定一次调用是否需要 ask？** 策略生产者：返回 `permissionDecision: ask` 的钩子、任何 `tools/pre-execute` 监听器、或沙箱升级门禁。seam 和桥只负责路由和应答；二者都不注入自己对「什么值得弹出提示」的判断。 - **用户关闭提示或轮次在 ask 进行中中止时会发生什么？** 关闭映射为 `cancelled` 并携带自己的拒绝文本。已中止的 signal 直接结算为 `cancelled` 而不派发；ask 进行中的中止丢弃迟到的应答。当两个审计追加都提交时，任一路径都记录恰好一对事件，绝不会两对。 - **如果客户端以 harness 从未提供的选项应答呢？** 除已提供的 `allow_once` 之外的任何选项都映射为 `rejected`——来自不合规客户端的未知 optionId 永远不能授权。 -- **subagent 的审批如何路由？** 没有应答者拥有的 agent 穿过整个 waterfall 委派并失败关闭——进程内 subagent 被刻意设计为不可应答。`subagent-acp` 的子侧自动应答是独立的；将子 agent 的 ask 路由到父控制器已延后（§ 延后）。 +- **subagent 的审批如何路由？** 没有应答者拥有的 agent 穿过整个 waterfall 委派并失败关闭——进程内 subagent 被刻意设计为不可应答。`'never'` 父级会把该覆盖项预置到每个进程内子 agent 的日志中（[决策](2026-07-25-subagent-policy-inheritance.md)），因此子 agent 一开始就会得知，而不是向空的 waterfall 发出 ask。`subagent-acp` 的子侧自动应答是独立的；将子 agent 的 ask 路由到父控制器已延后（§ 延后）。 - **`policy: 'never'` 在运行时实际改变了什么？** 服务在派发任何应答者之前，将该会话的每次 ask 解析为 `rejected`（在服务内部，因此没有注册顺序能绕过它）；系统提示词声明该策略；切换在边界处被叙述；每次成功的自动拒绝都会记录审计对。 - **热重载或应答者在会话中途卸载时会发生什么？** 应答者随其拥有的 fiber 一起 dispose，因此下一次 ask 降级为 `unavailable` 而非挂在死通道上；重新挂载会重新注册应答者，无需追赶状态。 - **客户端从哪里获得审批上下文？** 请求携带精确的 `callId` 和发起方的人类可读 `reason`；通道适配器可自行关联更丰富的工具调用状态，而无需在审批 seam 中重复携带参数。 diff --git a/.agents/notes/implemented/feature/2026-07-06-sandbox.i18n.yaml b/.agents/notes/implemented/feature/2026-07-06-sandbox.i18n.yaml index 952027b6e3..9311c6cac7 100644 --- a/.agents/notes/implemented/feature/2026-07-06-sandbox.i18n.yaml +++ b/.agents/notes/implemented/feature/2026-07-06-sandbox.i18n.yaml @@ -2,5 +2,5 @@ # side as of the last confirmed-consistent state. Both languages carry equal authority; # after editing either side, bring the other along and re-record with: # pnpm run verify-translation-pairing --write .agents/notes/implemented/feature/2026-07-06-sandbox.md -2026-07-06-sandbox.md: c3c61ed4539bcbca359f84f3dcc020e0bd41ae79 -2026-07-06-sandbox.zh.md: 39bf92aa20c697d2ad5f4c0926caeff0a0b60a1d +2026-07-06-sandbox.md: 42b78ad8341dd52c4dd146a2207a5ae909d28f1e +2026-07-06-sandbox.zh.md: dfa3349e4d74d6f2c4944414c25fe3726d4a9b5a diff --git a/.agents/notes/implemented/feature/2026-07-06-sandbox.md b/.agents/notes/implemented/feature/2026-07-06-sandbox.md index c3c61ed453..42b78ad834 100644 --- a/.agents/notes/implemented/feature/2026-07-06-sandbox.md +++ b/.agents/notes/implemented/feature/2026-07-06-sandbox.md @@ -89,10 +89,10 @@ Left open: what a durable grant's scope identity is beyond the sandbox mode — #### Per-session modes: the session log as the store ``` -effective(session) = findLast(the session's own knob events)?.value ?? the composition-config default +effective(session) = findLast(the session's knob events)?.value ?? the composition-config default ``` -The default is composition config (`cordis.yml`) — operator-owned, process-wide. A runtime switch is a session-scoped override recorded as one log-only event in that session's own log. Restart immunity (resuming a session replays its log, so overrides come back with zero catch-up machinery) and multi-session isolation both fall out by construction, and no external config store exists anywhere. +The default is composition config (`cordis.yml`) — operator-owned, process-wide. A runtime switch is a session-scoped override recorded as one log-only event in that session's log. Restart immunity and multi-session isolation follow from replay, with no external config store. The in-process subagent driver snapshots a parent's explicit override at delegation and seeds a source-tagged event after the child's optional fork prefix, so delegation cannot fall back to a wider default ([decision](2026-07-25-subagent-policy-inheritance.md)). **One event per knob, owned by its domain** — the merge-extensible `SessionEventMap` idiom every existing event family already follows (`approval/*` in `dsh-user-approval`, `hook/*` in the hooks packages): diff --git a/.agents/notes/implemented/feature/2026-07-06-sandbox.zh.md b/.agents/notes/implemented/feature/2026-07-06-sandbox.zh.md index 39bf92aa20..dfa3349e4d 100644 --- a/.agents/notes/implemented/feature/2026-07-06-sandbox.zh.md +++ b/.agents/notes/implemented/feature/2026-07-06-sandbox.zh.md @@ -89,10 +89,10 @@ Landlock launcher 源码和包工作区位于 `native/landlock-run`，与 harnes #### 按会话模式：会话日志即存储 ``` -effective(session) = findLast(the session's own knob events)?.value ?? the composition-config default +effective(session) = findLast(the session's knob events)?.value ?? the composition-config default ``` -默认值是组合配置（`cordis.yml`）——运维人员拥有，进程范围。运行时切换是会话范围的覆盖，记录为该会话自身日志中的一条仅日志事件。重启免疫（恢复会话时回放其日志，覆盖自然恢复，无需追赶机制）和多会话隔离都是构造性的自然结果，且不存在任何外部配置存储。 +默认值是组合配置（`cordis.yml`）——由运维人员拥有、作用于整个进程。运行时切换是会话范围的覆盖，以一条仅日志事件记录在该会话的日志中。重启免疫与多会话隔离由回放自然保证，且不存在任何外部配置存储。进程内 subagent 驱动器在委派时对父级的显式覆盖项获取快照，并在子 agent 可选的 fork 前缀之后预置一条带来源标记的事件，因此委派无法回退到更宽的默认值（[决策](2026-07-25-subagent-policy-inheritance.md)）。 **每个旋钮一种事件，由其领域拥有**——这是每个既有事件族已遵循的可合并扩展 `SessionEventMap` 惯用法（`dsh-user-approval` 中的 `approval/*`、hooks 包中的 `hook/*`）： diff --git a/.agents/notes/implemented/feature/2026-07-20-dsh-cli-personal-config.i18n.yaml b/.agents/notes/implemented/feature/2026-07-20-dsh-cli-personal-config.i18n.yaml index 76dd488060..401d02aeac 100644 --- a/.agents/notes/implemented/feature/2026-07-20-dsh-cli-personal-config.i18n.yaml +++ b/.agents/notes/implemented/feature/2026-07-20-dsh-cli-personal-config.i18n.yaml @@ -1,6 +1,6 @@ # Bilingual-pair consistency record (docs/i18n/README.md): the git blob hash of each # side as of the last confirmed-consistent state. Both languages carry equal authority; # after editing either side, bring the other along and re-record with: -# pnpm run verify-translation-pairing --write -2026-07-20-dsh-cli-personal-config.md: 9525aa811d792a918f03a52c21bc273e92fb8be7 -2026-07-20-dsh-cli-personal-config.zh.md: f21d4b1f22b3a3807b6b4155969282343f6048f5 +# pnpm run verify-translation-pairing --write .agents/notes/implemented/feature/2026-07-20-dsh-cli-personal-config.md +2026-07-20-dsh-cli-personal-config.md: cc965438214b68078647596af5a28fd666e7bd95 +2026-07-20-dsh-cli-personal-config.zh.md: 02f9c578061e1c25044c377c3ec8f80594275ae7 diff --git a/.agents/notes/implemented/feature/2026-07-20-dsh-cli-personal-config.md b/.agents/notes/implemented/feature/2026-07-20-dsh-cli-personal-config.md index 9525aa811d..cc96543821 100644 --- a/.agents/notes/implemented/feature/2026-07-20-dsh-cli-personal-config.md +++ b/.agents/notes/implemented/feature/2026-07-20-dsh-cli-personal-config.md @@ -12,7 +12,7 @@ A developer's own preferences — which provider and model the TUI uses, persona Two coupled pieces, aligned with the `apps/` assembly tier proposed by the `dsh web` PR (#443): -**The `dsh` CLI (`apps/cli`, npm name `@deepseek-ai/dsh`).** `apps/*` joins the workspaces as the product-assembly tier over `packages/*` libraries. The bin's dispatch reserves `web` and `-p`/`--prompt` for PR #443 (they exit with a pointer) so the two branches merge as a near-union; everything else runs the default surface: the interactive TUI, booting the shipped `examples/tui-agent/cordis.yml` (or an explicit config argument) with the invoking directory as the workspace. The committed `bin/dsh` launcher resolves the checkout through its own real path and runs the bin **from source** via the repo's tsx, so `ln -sf "$(pwd)/bin/dsh" ~/.local/bin/dsh` installs a command that always executes the current working tree. `pnpm run demo:tui` runs the same entry. +**The `dsh` CLI (`apps/cli`, npm name `@deepseek-ai/dsh`).** `apps/*` joins the workspaces as the product-assembly tier over `packages/*` libraries. The bin's dispatch reserves `web` and `-p`/`--prompt` for PR #443 (they exit with a pointer) so the two branches merge as a near-union; everything else runs the default surface: the interactive TUI, booting the shipped `examples/tui-agent/cordis.yml` (or an explicit config argument) with the invoking directory as the workspace. The committed `bin/dsh` launcher resolves the checkout through its own real path and runs the bin **from source** through Node's native TypeScript transform plus the app-owned tsconfig-paths loader, so `ln -sf "$(pwd)/bin/dsh" ~/.local/bin/dsh` installs a command that always executes the current working tree. `pnpm run demo:tui` runs the same entry. **Personal config (`dsh-app-boot`).** The personal overlay lives in the Harness home — `$DSH_HOME`, else `~/.dsh` — resolved by the shared [`resolveDshHome`](../architecture/2026-07-24-single-harness-home-resolver.md) (`@deepseek-ai/dsh-paths`), the same single root skills and AGENTS.md resolve against. The dsh TUI surface consumes its two optional files; the demo bins boot their committed trees verbatim: diff --git a/.agents/notes/implemented/feature/2026-07-20-dsh-cli-personal-config.zh.md b/.agents/notes/implemented/feature/2026-07-20-dsh-cli-personal-config.zh.md index f21d4b1f22..02f9c57806 100644 --- a/.agents/notes/implemented/feature/2026-07-20-dsh-cli-personal-config.zh.md +++ b/.agents/notes/implemented/feature/2026-07-20-dsh-cli-personal-config.zh.md @@ -12,7 +12,7 @@ Status: implemented 两个耦合的部分，与 `dsh web` PR（#443）提出的 `apps/` 装配层对齐： -**`dsh` CLI（`apps/cli`，npm 名 `@deepseek-ai/dsh`）。** `apps/*` 作为 `packages/*` 库之上的产品装配层加入 workspaces。bin 的分发把 `web` 和 `-p`/`--prompt` 保留给 PR #443（它们以指引退出），使两个分支能以接近并集的方式合并；其余一切都运行默认表面：交互式 TUI，加载随仓库提供的 `examples/tui-agent/cordis.yml`（或显式的配置参数），并以调用目录为工作区。已提交的 `bin/dsh` 启动器通过自身真实路径解析 checkout，用仓库的 tsx **从源码**运行该 bin，因此 `ln -sf "$(pwd)/bin/dsh" ~/.local/bin/dsh` 安装的命令永远执行当前工作树。`pnpm run demo:tui` 运行同一入口。 +**`dsh` CLI（`apps/cli`，npm 名 `@deepseek-ai/dsh`）。** `apps/*` 作为 `packages/*` 库之上的产品装配层加入 workspaces。bin 的分发把 `web` 和 `-p`/`--prompt` 保留给 PR #443（它们以指引退出），使两个分支能以接近并集的方式合并；其余一切都运行默认表面：交互式 TUI，加载随仓库提供的 `examples/tui-agent/cordis.yml`（或显式的配置参数），并以调用目录为工作区。已提交的 `bin/dsh` 启动器通过自身真实路径解析 checkout，通过 Node 的原生 TypeScript 转换和应用自身持有的 tsconfig-paths loader **从源码**运行该 bin，因此 `ln -sf "$(pwd)/bin/dsh" ~/.local/bin/dsh` 安装的命令永远执行当前工作树。`pnpm run demo:tui` 运行同一入口。 **个人配置（`dsh-app-boot`）。** 个人 overlay 存放在 Harness home——`$DSH_HOME`，否则 `~/.dsh`——由共享的 [`resolveDshHome`](../architecture/2026-07-24-single-harness-home-resolver.md)（`@deepseek-ai/dsh-paths`）解析，与 skills、AGENTS.md 解析所依据的单一根目录相同。dsh 的 TUI 表面消费其中两个可选文件；各示例 bin 仍然逐字节按已提交的配置树启动： diff --git a/.agents/notes/implemented/feature/2026-07-22-docked-web-goal-bar.i18n.yaml b/.agents/notes/implemented/feature/2026-07-22-docked-web-goal-bar.i18n.yaml new file mode 100644 index 0000000000..48a75f3449 --- /dev/null +++ b/.agents/notes/implemented/feature/2026-07-22-docked-web-goal-bar.i18n.yaml @@ -0,0 +1,6 @@ +# Bilingual-pair consistency record (docs/i18n/README.md): the git blob hash of each +# side as of the last confirmed-consistent state. Both languages carry equal authority; +# after editing either side, bring the other along and re-record with: +# pnpm run verify-translation-pairing --write .agents/notes/implemented/feature/2026-07-22-docked-web-goal-bar.md +2026-07-22-docked-web-goal-bar.md: 52a7d223ce3522c5ba977b1126dcd63bd2f6366f +2026-07-22-docked-web-goal-bar.zh.md: e4842a03ccb8a29b35c7af0c03c51b1324b6ab36 diff --git a/.agents/notes/implemented/feature/2026-07-22-docked-web-goal-bar.md b/.agents/notes/implemented/feature/2026-07-22-docked-web-goal-bar.md new file mode 100644 index 0000000000..52a7d223ce --- /dev/null +++ b/.agents/notes/implemented/feature/2026-07-22-docked-web-goal-bar.md @@ -0,0 +1,39 @@ +# Agent Note: Docked web goal bar + +Status: implemented + +English | [中文](2026-07-22-docked-web-goal-bar.zh.md) + +## Problem + +The web UI had no goal surface at all: the goal stack shipped with model tools, the TUI/ACP adapters, and the `/goal` command, but the browser client exposed none of it — no runtime verbs, no indicator. This change introduces the client goal verbs (runtime session methods over RPC) and the first goal UI together. Placement follows the redesign's premise that goal presence belongs to the composer's context: the goal is a property of the work the user is about to prompt, so its indicator docks directly above the message composer as a rounded-top strip tucked under the composer card's top edge. The mock keeps only a sparkle, a phase word ("Ongoing/Paused/Blocked Goal"), the truncated objective, and edit/clear icon actions, with resume appearing only on a paused goal. + +## Decision + +`GoalBar` (`packages/client/ui-goal/src/client/GoalBar.tsx`) is a new props-driven, self-contained component; `ConversationRoot` mounts it immediately before the composer `InputBar`. The strip's CSS mirrors the composer's horizontal geometry (32px side padding, 776px centered cap) plus the mock's 12px inset, and a -10px bottom margin eats InputBar's 8px top padding and tucks its square bottom edge 2px under the composer card's top edge. All strip states share one fixed 38px height so switching between them never resizes it. Loading (`goal === undefined`), absent (`goal === null`), and `phase === 'complete'` render nothing — a completed goal is history, not chrome. + +Visibility drives the label and actions: active shows "Ongoing Goal" with edit/clear; paused shows "Paused Goal" and adds a resume icon button; blocked shows "Blocked Goal" and carries `blockedReason.message` as the strip's `title` tooltip. Goal creation lives on the `/goal` command, not in the bar. The pencil swaps the strip for an inline edit form prefilled with the current objective: Enter or the check button saves through `GoalBarActions.onEdit(objective)`, Esc cancels, and an all-whitespace objective keeps save disabled. The form closes only when the edit succeeds; a failure preserves the draft and displays the error in the bar. Resume and clear failures are displayed there as well. Clear otherwise calls `onClear` directly with no confirmation — a clear keeps a durable tombstone, so nothing is unrecoverable. An effect keyed on the goal's id drops the edit form when the goal's identity changes, so a surviving draft can never be written over the goal that replaced it. + +`GoalBarActions` lives in the contract layer (`contract/slots.ts`, next to the `ConversationInjected.goalActions` slot it feeds) and carries exactly the rendered verbs: `onEdit`/`onResume`/`onClear`. Each callback asynchronously returns an explicit success/failure result so `GoalBar` owns its transitions and error display. `apply.ts` wires them to the runtime session methods; the runtime session resolves the current goal's compare-and-set ref internally, so the UI passes no ref. + +The runtime session gains the goal surface the strip (and future UI) needs: `fetchGoal` populates the snapshot on open, and a live `context/message` carrying `goal/change` meta triggers a coalesced refetch — concurrent triggers share the in-flight `goal.get`, while a trigger received during that read schedules one coalesced trailing read so independently ordered notifications and GET responses cannot leave stale state. Window replays never refetch, and matching the meta kind (rather than a goal key) also catches clear tombstones written by other clients. The six mutation verbs fold transport failures into `{ ok: false }` results like every sibling session method, and a get result older than a mutation response that landed mid-flight is dropped. + +The strip's background is `--dsw-alias-interactive-bg-hover` rather than the mock's literal `#F5F6F7`: the translucent hover gray resolves to that value over the white light-theme base and lifts the strip off the composer card in dark mode, where a static light token would sink. All colors are `--dsw-*` tokens. + +## Testing + +`packages/client/ui-goal/tests/goalbar.spec.tsx` pins the behavior through props alone: loading/absent/complete render nothing, the active strip renders label/objective and fires clear, the edit form prefills, rejects empty, saves on Enter, cancels on Esc, and resets when the goal's identity changes, the paused strip fires resume, and the blocked strip exposes the reason tooltip. Component failure-path cases prove that a failed edit preserves its draft and that edit/resume/clear errors remain visible in the bar. The skeleton specs mount `ConversationRoot` with and without `goalActions`; the undefined case is seeded with an active goal, so the missing gate — not the missing goal — is what hides the strip. Runtime session specs pin the folded-error results, the live-only in-flight-plus-trailing refetch, and the stale-read guard. A keyless real-browser smoke boots the assembled application through `boot → RPC → runtime → GoalBar` and records an inline snapshot of the rendered label, objective, and actions. + +## Alternatives considered + +- **Put the strip in the session header** — rejected because the redesign's premise is that goal presence belongs to the composer's context; a header strip cannot dock into the composer card. +- **Render a "Loading goal…" placeholder for `undefined`** — rejected: the strip would flash and collapse on every session open, chrome noise for a sub-second state. +- **Include an inline create affordance when no goal is set** — rejected after implementation review: goal creation lives on the `/goal` command, matching the pattern where the model creates goals on request; the bar is a status indicator, not a creation surface. +- **Carry the full verb set (`onPause`/`onComplete`) in `GoalBarActions`** — rejected as speculative generality: no consumer calls them, so the interface carries only the rendered verbs. + +## Consequences + +- Goal presence in the web UI is a composer-docked strip: sparkle, phase label, truncated objective, and edit/clear (plus resume when paused) — the browser client's first goal surface. +- The runtime session exposes the goal verbs over RPC with folded transport errors, and refreshes the snapshot's goal on open and on live goal-change meta (coalesced, guarded against stale reads). +- Objective editing is reachable from the UI for the first time, through `goal.edit` with the runtime-owned ref; pause/complete remain available to other surfaces (`/goal`, model tools). +- `goal === null` renders nothing; the composer carries no persistent create affordance — creation is the `/goal` command's job. diff --git a/.agents/notes/implemented/feature/2026-07-22-docked-web-goal-bar.zh.md b/.agents/notes/implemented/feature/2026-07-22-docked-web-goal-bar.zh.md new file mode 100644 index 0000000000..e4842a03cc --- /dev/null +++ b/.agents/notes/implemented/feature/2026-07-22-docked-web-goal-bar.zh.md @@ -0,0 +1,39 @@ +# Agent Note: 停靠式 Web 目标条 + +Status: implemented + +[English](2026-07-22-docked-web-goal-bar.md) | 中文 + +## 问题 + +Web UI 此前没有任何目标相关的界面：目标栈已随模型工具、TUI/ACP 适配器和 `/goal` 命令交付，但浏览器客户端完全不接触它——既没有运行时动词，也没有指示器。本变更同时引入客户端目标动词（基于 RPC 的运行时会话方法）和第一个目标 UI。摆放位置遵循重新设计的前提：目标的存在感属于输入框的上下文——目标是用户即将提交的工作的属性，因此它的指示器停靠在消息输入框正上方，呈现为一条圆角顶部的横条，收进输入框卡片顶边之下。设计稿只保留一个闪光图标、一个阶段词（"Ongoing/Paused/Blocked Goal"）、截断后的目标内容，以及编辑／清除图标操作，恢复按钮仅在目标暂停时出现。 + +## 决策 + +`GoalBar`（`packages/client/ui-goal/src/client/GoalBar.tsx`）是一个新的、由 props 驱动的自包含组件；`ConversationRoot` 将它挂载在输入框 `InputBar` 紧上方。横条的 CSS 对齐输入框的水平几何（两侧 32px 内边距、776px 居中上限），再加上设计稿的 12px 内缩，并用 -10px 的下外边距吃掉 InputBar 的 8px 上内边距，使它方形的底边收进输入框卡片顶边之下 2px。横条的所有状态共享固定的 38px 高度，状态切换不会引起尺寸变化。加载中（`goal === undefined`）、无目标（`goal === null`）和 `phase === 'complete'` 时不渲染任何内容：已完成的目标是历史记录，不是常驻界面元素。 + +可见性决定标签和操作：active 状态显示 "Ongoing Goal" 并提供编辑／清除；paused 状态显示 "Paused Goal"，并增加一个恢复图标按钮；blocked 状态显示 "Blocked Goal"，并把 `blockedReason.message` 作为横条的 `title` 悬浮提示。创建目标的入口在 `/goal` 命令上，不在横条里。铅笔图标把横条切换为内联编辑表单，预填当前目标内容：Enter 或勾选按钮通过 `GoalBarActions.onEdit(objective)` 保存，Esc 取消，目标内容全为空白字符时保存按钮保持禁用。编辑成功后表单才会关闭；编辑失败时保留草稿，并在横条中显示错误。恢复和清除失败也显示在横条中。除此之外，清除直接调用 `onClear`，不做确认——清除会保留 durable 墓碑，没有不可恢复的损失。一个以目标 id 为键的 effect 会在目标身份变化时丢弃编辑表单，因此存留的草稿绝不可能覆盖掉替换它的新目标。 + +`GoalBarActions` 位于 contract 层（`contract/slots.ts`，紧挨它所喂给的 `ConversationInjected.goalActions` 槽位），只携带实际渲染的动词：`onEdit`/`onResume`/`onClear`。每个回调都会异步返回显式成功／失败结果，因此 `GoalBar` 自行负责界面转换和错误显示。`apply.ts` 把它们接到运行时会话方法上；运行时会话在内部解析当前目标的 compare-and-set ref，因此 UI 不传 ref。 + +运行时会话获得了横条（以及未来 UI）所需的目标表面：`fetchGoal` 在打开时填充快照；携带 `goal/change` 元数据的 live `context/message` 触发合并重新拉取——并发触发器共享正在执行的 `goal.get`，读取期间收到的触发器会安排一次合并后的尾随读取，避免彼此独立排序的通知和 GET 响应留下陈旧状态。窗口重放绝不触发重新拉取，且匹配元数据 kind（而不是 goal 键）还能捕获其他客户端写入的清除墓碑。六个变更动词与所有同类会话方法一样，把传输层失败折叠为 `{ ok: false }` 结果；比在拉取途中落地的变更响应更旧的 get 结果会被丢弃。 + +横条的背景色用 `--dsw-alias-interactive-bg-hover`，而不是设计稿里的字面值 `#F5F6F7`：这个半透明的悬浮灰在浅色主题的白色底上正好解析为该值，而在深色模式下能把横条从输入框卡片上衬托出来，静态的浅色 token 在深色模式下会沉进去。所有颜色都是 `--dsw-*` token。 + +## 测试 + +`packages/client/ui-goal/tests/goalbar.spec.tsx` 仅通过 props 固定这些行为：加载中／无目标／已完成时不渲染；active 横条渲染标签和目标内容并触发清除；编辑表单预填内容、拒绝空值、按 Enter 保存、按 Esc 取消，并在目标身份变化时重置；paused 横条触发恢复；blocked 横条暴露原因悬浮提示。组件失败路径用例证明编辑失败时保留草稿，并且编辑／恢复／清除错误持续显示在横条中。skeleton 规格测试分别挂载带与不带 `goalActions` 的 `ConversationRoot`；未定义的情形预置了一个 active 目标，因此隐藏横条的是缺失的挂载门，而不是缺失的目标。运行时会话规格测试固定了折叠错误结果、仅 live 的执行中读取加尾随读取，以及陈旧读取守卫。一个无密钥真实浏览器冒烟测试通过 `boot → RPC → runtime → GoalBar` 启动组装后的应用，并以内联快照记录渲染出的标签、目标内容和操作。 + +## 考虑过的替代方案 + +- **把横条放在会话头部**：不予采纳，因为重新设计的前提是目标的存在感属于输入框的上下文；放在头部的横条无法停靠进输入框卡片。 +- **为 `undefined` 渲染 "Loading goal…" 占位**：不予采纳，每次打开会话横条都会闪现再坍缩，对一个不到一秒的状态来说只是界面噪音。 +- **未设置目标时在横条内提供内联创建入口**：实现评审后不予采纳，创建目标的职责在 `/goal` 命令上，与模型按请求创建目标的模式一致；横条是状态指示器，不是创建入口。 +- **在 `GoalBarActions` 中携带完整动词集合（`onPause`/`onComplete`）**：作为投机性泛化不予采纳，没有消费方调用它们，接口只携带实际渲染的动词。 + +## 后果 + +- Web UI 中目标的存在形式是停靠在输入框上方的横条：闪光图标、阶段标签、截断的目标内容，以及编辑／清除（暂停时另有恢复）——这是浏览器客户端的第一个目标界面。 +- 运行时会话通过 RPC 暴露目标动词并折叠传输层错误，且在打开时和 live 目标变更元数据到达时刷新快照中的目标（合并拉取，带陈旧读取守卫）。 +- 目标内容首次可以从 UI 编辑，经由 `goal.edit`，ref 由运行时持有；暂停／完成对其他界面（`/goal`、模型工具）照常可用。 +- `goal === null` 时不渲染任何内容；输入框不提供常驻的创建入口，创建是 `/goal` 命令的职责。 diff --git a/.agents/notes/implemented/feature/2026-07-25-subagent-policy-inheritance.i18n.yaml b/.agents/notes/implemented/feature/2026-07-25-subagent-policy-inheritance.i18n.yaml new file mode 100644 index 0000000000..ab957282ba --- /dev/null +++ b/.agents/notes/implemented/feature/2026-07-25-subagent-policy-inheritance.i18n.yaml @@ -0,0 +1,6 @@ +# Bilingual-pair consistency record (docs/i18n/README.md): the git blob hash of each +# side as of the last confirmed-consistent state. Both languages carry equal authority; +# after editing either side, bring the other along and re-record with: +# pnpm run verify-translation-pairing --write .agents/notes/implemented/feature/2026-07-25-subagent-policy-inheritance.md +2026-07-25-subagent-policy-inheritance.md: aeff83795eedead9c75de6bbb74c1da1945092ca +2026-07-25-subagent-policy-inheritance.zh.md: 7a09fd972f53b46655df93ddf9625455f2077460 diff --git a/.agents/notes/implemented/feature/2026-07-25-subagent-policy-inheritance.md b/.agents/notes/implemented/feature/2026-07-25-subagent-policy-inheritance.md new file mode 100644 index 0000000000..aeff83795e --- /dev/null +++ b/.agents/notes/implemented/feature/2026-07-25-subagent-policy-inheritance.md @@ -0,0 +1,36 @@ +# Agent Note: In-process subagent policy inheritance — the child starts under the parent's sandbox and approval overrides + +Status: implemented + +English | [中文](2026-07-25-subagent-policy-inheritance.zh.md) + +## Problem + +Sandbox and approval overrides are per-session log folds. An in-process subagent gets a new session, so a spawn child once fell back to deployment defaults and a fork child saw only switches inside its completed-turn prefix. Delegation could therefore widen a parent that had switched to `read-only`, or turn a parent's unattended `'never'` approval stance back into prompting behavior. + +## Decision + +The shared in-process driver snapshots `sandboxPolicy.overrideOf(parent.session)` and `approval.overrideOf(parent.session)` before its first await. A later parent switch belongs to the parent's future; cancel-and-redelegate takes a new snapshot. Both services are optional, and only explicit session overrides are copied, never deployment defaults or one-shot grants. + +Each captured value becomes a source-tagged `sandbox/mode` or `approval/policy` event appended during the child factory's unpublished setup. The session constructor has already fixed `Session.firstLiveSeq` at the fork-prefix length, so the inherited facts follow fork history, reach telemetry when the child is announced, and leave `SessionHeader.seedLength` at the prefix length. Existing last-event-wins folds therefore make the delegation snapshot beat stale fork history and let a later child switch beat the snapshot. A grandchild folds its parent's logged state, so the rule composes without another inheritance mechanism. + +Ordinary session appends validate the inherited events before publication, and persistence captures the complete unpublished log when the session is announced. Any materialized child log therefore stores the inherited events with its first batch; there is no second policy store, schema field, or query index. The `source: 'delegation'` marker lets approval narration distinguish inheritance from a child-side user switch. + +### What a blocked child experiences + +A confined child gets the ordinary denial marker. No answerer currently owns an in-process child, so an escalation request fails closed and the child reports upward; a controller-owned parent may widen its own session and delegate again. An inherited `'never'` policy tells the child not to request escalation in its first system prompt. + +## Alternatives considered + +- **Generic `SessionHeader` policy fields** — rejected: they duplicate an event-sourced fact in metadata and require propagation through core session types, persistence backends, query indexes, collision identity, and every policy consumer. Unpublished setup events have the required ordering and reuse the existing durable store. +- **Combining new policy facts with constructor history** — rejected: `Session.firstLiveSeq` classifies the complete constructor seed as replayed history, so telemetry would skip child-only facts. Unpublished setup keeps history and new facts on their existing sides of that boundary without another session option. +- **A first-prompt listener** — rejected: it introduces listener ordering and a later timing boundary even though the creation transaction already permits log appends before publication. +- **Copying deployment defaults** — rejected: defaults remain operator-owned and may change; an unswitched parent stamps nothing, so its child follows the current deployment. +- **Live resolution walking `parentSession` at each call** — rejected: it breaks the "two sessions never see each other's state" isolation invariant, requires the parent session to stay loaded for the child's lifetime, and makes a mid-run parent switch retroactively change a running child. Snapshot-at-delegation is the semantic: the child keeps the policy it was handed; cancel-and-respawn picks up a tightening. +- **Forcing `'never'` or routing asks to the root controller** — rejected as inheritance behavior. A forced value forecloses a future child answerer; parent routing needs parent-chain ownership and the spawning `callId`, and remains deferred in [the approval-seam Agent Note](2026-07-06-approval-seam.md). + +## Consequences + +- Spawn, fork, and nested in-process children retain a parent's explicit sandbox and approval overrides. The focused suite proves real filesystem denial, stale-fork precedence, delegation-time capture, the live-event boundary, default omission, and context disposal. +- The keyless headless snapshot is the assembled regression: only the parent is `read-only`, the deployment default is `workspace-write`, and the child's persisted event plus denied disk write both fail if capture is removed. +- Each delegation adds at most two log-only events. `dsh-subagent-inprocess` has optional peer types for the two policy services; compositions without either service behave unchanged. Out-of-process children retain their own deployment policy, and a running child does not follow later parent switches. diff --git a/.agents/notes/implemented/feature/2026-07-25-subagent-policy-inheritance.zh.md b/.agents/notes/implemented/feature/2026-07-25-subagent-policy-inheritance.zh.md new file mode 100644 index 0000000000..7a09fd972f --- /dev/null +++ b/.agents/notes/implemented/feature/2026-07-25-subagent-policy-inheritance.zh.md @@ -0,0 +1,36 @@ +# Agent Note: 进程内 subagent 策略继承——子 agent 在父级的沙箱与审批覆盖项下启动 + +Status: implemented + +[English](2026-07-25-subagent-policy-inheritance.md) | 中文 + +## 问题 + +沙箱与审批覆盖项都是按会话的日志折叠。进程内 subagent 会获得一个新会话，因此 spawn 子 agent（智能体）过去会回退到部署默认值，fork 子 agent 则只能看到其已完成轮次前缀中的切换。因此，委派可能放宽已经切换到 `read-only` 的父级，或让父级无人值守的 `'never'` 审批立场重新变成会发起提示的行为。 + +## 决策 + +共享的进程内驱动器在第一次 await 之前对 `sandboxPolicy.overrideOf(parent.session)` 和 `approval.overrideOf(parent.session)` 获取快照。父级后续的切换属于父级的未来；取消后重新委派会取得新快照。这两个服务均为可选，仅复制显式会话覆盖项，绝不复制部署默认值或一次性授权。 + +每个捕获值都会成为子 agent 工厂在未发布设置阶段追加的一条带来源标记的 `sandbox/mode` 或 `approval/policy` 事件。会话构造函数已将 `Session.firstLiveSeq` 固定为 fork 前缀的长度，因此继承事实会排在 fork 历史之后，在子 agent 公布时进入遥测，同时让 `SessionHeader.seedLength` 保持为此前缀的长度。因此，既有的末事件胜出折叠会让委派快照压过陈旧的 fork 历史，并让子 agent 后续的切换压过该快照。孙代 agent 会折叠其父级已记录的状态，因此无需另一套继承机制即可组合此规则。 + +普通的会话追加会在发布前校验继承事件，持久化层则在会话公布时捕获完整的未发布日志。因此，任何已物化的子 agent 日志都会在首批数据中存下继承事件；不存在第二套策略存储、schema 字段或查询索引。`source: 'delegation'` 标记让审批叙述能够区分继承与子 agent 侧的用户切换。 + +### 被拦住的子 agent 会经历什么 + +受限子 agent 会得到普通拒绝标记。目前没有应答器认领进程内子 agent，因此升级请求会失败关闭，由子 agent 向上汇报；由控制器持有的父 agent 可以放宽自己的会话后重新委派。继承的 `'never'` 策略会在第一份系统提示词中告知子 agent 不要请求升级。 + +## 考虑过的替代方案 + +- **通用的 `SessionHeader` 策略字段**：不予采纳。它们会在元数据中复制一项事件溯源事实，并要求贯穿核心会话类型、持久化后端、查询索引、碰撞标识与每个策略消费方进行传播。未发布设置阶段的事件具备所需顺序，并复用现有持久化存储。 +- **将新策略事实与构造历史合并**：不予采纳。`Session.firstLiveSeq` 会把完整的构造种子归类为回放历史，因此遥测会跳过仅属于子 agent 的事实。未发布设置让历史与新事实留在该边界各自原有的一侧，无需再增加会话选项。 +- **首个提示词监听器**：不予采纳。尽管创建事务已经允许在发布前追加日志，它仍会引入监听器顺序与更晚的时序边界。 +- **复制部署默认值**：不予采纳。默认值仍由运维人员拥有且可能变化；未切换的父级不会盖章写入任何内容，因此其子 agent 跟随当前部署。 +- **每次调用时沿 `parentSession` 实时解析**：不予采纳。这会打破「两个会话永远看不到彼此状态」的隔离不变量，要求父会话在子 agent 的整个生命周期内保持加载，还会让父级在子 agent 运行途中做的切换追溯性地改变一个正在运行的子 agent。委派时快照才是本设计的语义：子 agent 保持它被交付时的策略；取消后重新 spawn 即可拿到收紧后的策略。 +- **强制使用 `'never'` 或把 ask 路由到根控制器**：不作为继承行为采纳。强制值会排除未来的子 agent 应答器；父级路由需要父链所有权与发起 spawn 的 `callId`，仍按[审批 seam Agent Note](2026-07-06-approval-seam.md) 所述延期。 + +## 后果 + +- spawn、fork 和嵌套的进程内子 agent 会保留父级显式的沙箱与审批覆盖项。聚焦测试套件证明真实文件系统拒绝、陈旧 fork 优先级、委派时捕获、实时事件边界、默认值省略与上下文释放。 +- 无密钥 headless 快照是组装后应用层面的回归测试：只有父级是 `read-only`，部署默认值是 `workspace-write`；若移除捕获，子 agent 的持久化事件与被拒的磁盘写入这两项检查都会失败。 +- 每次委派最多增加两条仅日志事件。`dsh-subagent-inprocess` 为两个策略服务提供可选 peer 类型；未组合任一服务的组合保持原有行为。进程外子 agent 仍采用自身的部署策略，正在运行的子 agent 不跟随父级后续切换。 diff --git a/.agents/notes/implemented/process/2026-06-16-pnpm-over-yarn.i18n.yaml b/.agents/notes/implemented/process/2026-06-16-pnpm-over-yarn.i18n.yaml index 754b592530..6703e7bc9d 100644 --- a/.agents/notes/implemented/process/2026-06-16-pnpm-over-yarn.i18n.yaml +++ b/.agents/notes/implemented/process/2026-06-16-pnpm-over-yarn.i18n.yaml @@ -2,5 +2,5 @@ # side as of the last confirmed-consistent state. Both languages carry equal authority; # after editing either side, bring the other along and re-record with: # pnpm run verify-translation-pairing --write .agents/notes/implemented/process/2026-06-16-pnpm-over-yarn.md -2026-06-16-pnpm-over-yarn.md: 918da1d056dfec73f78d849e08ad06d397e654c0 -2026-06-16-pnpm-over-yarn.zh.md: 4fbb357f689b4a960d0662cbefa5bec30f79bcc1 +2026-06-16-pnpm-over-yarn.md: 30b34c65fdea94b20dec4d627a0fca40de760fd1 +2026-06-16-pnpm-over-yarn.zh.md: eb13890b7e7051301874b9273966771328b065a3 diff --git a/.agents/notes/implemented/process/2026-06-16-pnpm-over-yarn.md b/.agents/notes/implemented/process/2026-06-16-pnpm-over-yarn.md index 918da1d056..30b34c65fd 100644 --- a/.agents/notes/implemented/process/2026-06-16-pnpm-over-yarn.md +++ b/.agents/notes/implemented/process/2026-06-16-pnpm-over-yarn.md @@ -8,7 +8,7 @@ English | [中文](2026-06-16-pnpm-over-yarn.zh.md) The repo shipped on **Yarn 4** with the `node-modules` linker — a deliberately conservative choice that behaves like npm's flat layout while giving us Yarn's workspaces and `yarn constraints`. It worked. But Yarn 4's Plug'n'Play heritage makes the `node-modules` linker the off-the-beaten-path mode, and the broader JS ecosystem — tooling defaults, CI actions, Corepack examples, contributor familiarity — increasingly centers on pnpm. For a repo that is built primarily by agents and read by occasional human contributors, "the package manager most tools and people expect" has real value: fewer surprises, better-trodden failure paths, more copy-pasteable answers. -The switching cost is at its lowest right now. Nothing publishes from this repo yet (every package is `private: true`); dev/test/demo all run **unbuilt** via tsx, so the package manager only has to (a) resolve and link `node_modules`, (b) run the workspace scripts, and (c) enforce the workspace constraints. The one Yarn-specific asset is `yarn.config.cjs` (the `@yarnpkg/types` constraints engine), which is small and mechanical to re-express. This mirrors the reasoning in [the tsdown decision](../../archived/process/2026-06-11-tsdown-over-dumble.md): swap a load-bearing tool for the healthier-ecosystem option while the blast radius is still small. +The switching cost is at its lowest right now. Nothing publishes from this repo yet (every package is `private: true`); development, tests, and source-mode demos run through their declared TypeScript launchers, while artifact checks build explicitly. The package manager therefore only has to (a) resolve and link `node_modules`, (b) run the workspace scripts, and (c) enforce the workspace constraints. The one Yarn-specific asset is `yarn.config.cjs` (the `@yarnpkg/types` constraints engine), which is small and mechanical to re-express. This mirrors the reasoning in [the tsdown decision](../../archived/process/2026-06-11-tsdown-over-dumble.md): swap a load-bearing tool for the healthier-ecosystem option while the blast radius is still small. ## Decision diff --git a/.agents/notes/implemented/process/2026-06-16-pnpm-over-yarn.zh.md b/.agents/notes/implemented/process/2026-06-16-pnpm-over-yarn.zh.md index 4fbb357f68..eb13890b7e 100644 --- a/.agents/notes/implemented/process/2026-06-16-pnpm-over-yarn.zh.md +++ b/.agents/notes/implemented/process/2026-06-16-pnpm-over-yarn.zh.md @@ -8,7 +8,7 @@ Status: implemented 本仓库最初使用 **Yarn 4** 搭配 `node-modules` 链接器启动。这是一个刻意保守的选择：行为类似 npm 的扁平布局，同时享有 Yarn 的 workspaces 和 `yarn constraints`。它能正常工作。但 Yarn 4 源自 Plug'n'Play 的血统，使得 `node-modules` 链接器成为非主流模式；而更广泛的 JS 生态——工具默认值、CI action、Corepack 示例、贡献者的熟悉度——正日益以 pnpm 为中心。对于一个主要由 agent（智能体）构建、偶尔有人类贡献者阅读的仓库而言，「大多数工具和人所期望的包管理器」具有实际价值：更少的意外、更成熟的故障路径、更多可直接复用的解答。 -切换成本目前处于最低点。本仓库尚无任何包（package）发布（每个包都是 `private: true`）；开发/测试/演示全部通过 tsx **未构建**运行，因此包管理器只需做到：（a）解析并链接 `node_modules`，（b）运行 workspace 脚本，（c）强制执行 workspace 约束。唯一的 Yarn 特有资产是 `yarn.config.cjs`（`@yarnpkg/types` 约束引擎），体量小且可机械地重新表达。这与 [tsdown 决策](../../archived/process/2026-06-11-tsdown-over-dumble.md)的逻辑一致：在爆炸半径尚小时，将承重工具换为生态更健康的选项。 +切换成本目前处于最低点。本仓库尚无任何包（package）发布（每个包都是 `private: true`）；开发流程、测试和源码模式 demo 都通过各自声明的 TypeScript 启动器运行，产物检查则会显式构建。因此，包管理器只需做到：（a）解析并链接 `node_modules`，（b）运行 workspace 脚本，（c）强制执行 workspace 约束。唯一的 Yarn 特有资产是 `yarn.config.cjs`（`@yarnpkg/types` 约束引擎），体量小且可机械地重新表达。这与 [tsdown 决策](../../archived/process/2026-06-11-tsdown-over-dumble.md)的逻辑一致：在爆炸半径尚小时，将承重工具换为生态更健康的选项。 ## 决策 diff --git a/.agents/notes/implemented/process/2026-06-17-ts-build-config.i18n.yaml b/.agents/notes/implemented/process/2026-06-17-ts-build-config.i18n.yaml index 45ddcdfdae..4319bcee3e 100644 --- a/.agents/notes/implemented/process/2026-06-17-ts-build-config.i18n.yaml +++ b/.agents/notes/implemented/process/2026-06-17-ts-build-config.i18n.yaml @@ -1,6 +1,6 @@ # Bilingual-pair consistency record (docs/i18n/README.md): the git blob hash of each # side as of the last confirmed-consistent state. Both languages carry equal authority; # after editing either side, bring the other along and re-record with: -# pnpm run verify-translation-pairing --write -2026-06-17-ts-build-config.md: 5bdfc5e170f12cd95a68f443ab8d02b16db554f3 -2026-06-17-ts-build-config.zh.md: 9f74fb6be9c8e00a070e27a609edf8421a2b5ca6 +# pnpm run verify-translation-pairing --write .agents/notes/implemented/process/2026-06-17-ts-build-config.md +2026-06-17-ts-build-config.md: 20dd3d8d0a11e01397c36388903cd112a170f0bf +2026-06-17-ts-build-config.zh.md: 196ed3f2921a4be6f65cd56fc496993b7121f484 diff --git a/.agents/notes/implemented/process/2026-06-17-ts-build-config.md b/.agents/notes/implemented/process/2026-06-17-ts-build-config.md index 5bdfc5e170..20dd3d8d0a 100644 --- a/.agents/notes/implemented/process/2026-06-17-ts-build-config.md +++ b/.agents/notes/implemented/process/2026-06-17-ts-build-config.md @@ -62,7 +62,7 @@ pnpm run clean: tsx scripts/clean.ts ``` -`pnpm run demo:*` still runs `src` directly through tsx and root paths, without a compile step. +The source-mode demos run through their declared TypeScript launchers and the root paths map. The `dsh` TUI chain uses Node's native transform plus its app-owned paths loader, the Web demo builds its required artifacts before entering that same CLI source chain, and the other source demos continue to use tsx. ## Alternatives considered @@ -75,7 +75,7 @@ tsx scripts/clean.ts Build responsibilities are clearer: -- Each module under `packages//` and `vendor/*` has one local tsconfig for build, typecheck, and tools that run source directly, such as `tsx` and `vitest`. +- Each module under `packages//` and `vendor/*` has one local tsconfig for build, typecheck, and tools that run source directly, such as the `dsh` source loader, `tsx`, and `vitest`. - The `build` command drives the root solution graph. `tsc -b` owns the publishable per-module `.js` and `.d.ts` output, and the bundler owns only `lib/index.*`. - `lib/types/*.d.ts` and `.d.ts.map` are the publish declaration output. - `lib/types/*.d.ts` uses explicit `.ts` relative specifiers, which TypeScript's NodeNext/Node16 resolver maps to sibling `.d.ts` files. diff --git a/.agents/notes/implemented/process/2026-06-17-ts-build-config.zh.md b/.agents/notes/implemented/process/2026-06-17-ts-build-config.zh.md index 9f74fb6be9..196ed3f292 100644 --- a/.agents/notes/implemented/process/2026-06-17-ts-build-config.zh.md +++ b/.agents/notes/implemented/process/2026-06-17-ts-build-config.zh.md @@ -62,7 +62,7 @@ pnpm run clean: tsx scripts/clean.ts ``` -`pnpm run demo:*` 仍通过 tsx 和根路径直接运行 `src`，无需编译步骤。 +源码模式 demo 通过各自声明的 TypeScript 启动器和根路径映射运行。`dsh` TUI 链使用 Node 原生转换及应用自有的路径 loader，Web demo 在进入同一条 CLI 源码链路前先构建所需产物，其他源码 demo 继续使用 tsx。 ## 曾考虑的替代方案 @@ -75,7 +75,7 @@ tsx scripts/clean.ts 构建职责更加清晰： -- `packages//` 和 `vendor/*` 下的每个模块有一份本地 tsconfig，同时服务于构建、类型检查和直接运行源码的工具（如 `tsx` 和 `vitest`）。 +- `packages//` 和 `vendor/*` 下的每个模块有一份本地 tsconfig，同时服务于构建、类型检查和直接运行源码的工具（如 `dsh` 源码 loader、`tsx` 和 `vitest`）。 - `build` 命令驱动根 solution 图。`tsc -b` 负责可发布的逐模块 `.js` 和 `.d.ts` 输出，打包器仅负责 `lib/index.*`。 - `lib/types/*.d.ts` 和 `.d.ts.map` 是发布用的声明输出。 - `lib/types/*.d.ts` 使用显式 `.ts` 相对说明符，TypeScript 的 NodeNext/Node16 解析器会将其映射到同级的 `.d.ts` 文件。 diff --git a/.agents/notes/implemented/testing/2026-06-19-acp-snapshot-tests.i18n.yaml b/.agents/notes/implemented/testing/2026-06-19-acp-snapshot-tests.i18n.yaml index 48850b2a53..f126d6a85b 100644 --- a/.agents/notes/implemented/testing/2026-06-19-acp-snapshot-tests.i18n.yaml +++ b/.agents/notes/implemented/testing/2026-06-19-acp-snapshot-tests.i18n.yaml @@ -2,5 +2,5 @@ # side as of the last confirmed-consistent state. Both languages carry equal authority; # after editing either side, bring the other along and re-record with: # pnpm run verify-translation-pairing --write .agents/notes/implemented/testing/2026-06-19-acp-snapshot-tests.md -2026-06-19-acp-snapshot-tests.md: ba46682b3087e3d2ff4c52d2ad22f54b7dac31db -2026-06-19-acp-snapshot-tests.zh.md: 58195c47889edb31d5122116328928f916ad09a9 +2026-06-19-acp-snapshot-tests.md: e118ada58230fe31fbb2a6bffb83e5612757ab1f +2026-06-19-acp-snapshot-tests.zh.md: 5c239d2c2589708050cd6199231ec5047f225107 diff --git a/.agents/notes/implemented/testing/2026-06-19-acp-snapshot-tests.md b/.agents/notes/implemented/testing/2026-06-19-acp-snapshot-tests.md index ba46682b30..e118ada582 100644 --- a/.agents/notes/implemented/testing/2026-06-19-acp-snapshot-tests.md +++ b/.agents/notes/implemented/testing/2026-06-19-acp-snapshot-tests.md @@ -57,7 +57,7 @@ A snapshot run asserts **two** normalized surfaces, because the harness's extern The surfaces are complementary: stdout covers the minimal automation wire, while JSONL covers loop, tool, and boundary structure that the wire intentionally omits. -Normalization replaces session, cwd, protocol-id, timestamp, path, and process volatility while preserving deterministic sequence numbers. Scenarios constrain real bash use to stable commands. The stdout expected output remains wire-shaped JSONL and every raw line must parse as JSON. Vitest updates only the stdout expected output; normalized session equality never overwrites the replay fixture. +Normalization replaces session, cwd, protocol-id, timestamp, path, and process volatility while preserving deterministic sequence numbers. Record and refresh also store a generated workspace and its filesystem-resolved aliases as `{{cwd}}` in the replay fixture, so platform temp roots and random basenames do not affect recordings; authored temp paths and cwd values under an explicit `workspaceParent` remain literal. Scenarios constrain real bash use to stable commands. The stdout expected output remains wire-shaped JSONL and every raw line must parse as JSON. Ordinary Vitest snapshot updates write only the stdout expected output; the explicit `record` and `refresh` modes own replay-fixture writes. ### Isolation: normalization now, sandbox later diff --git a/.agents/notes/implemented/testing/2026-06-19-acp-snapshot-tests.zh.md b/.agents/notes/implemented/testing/2026-06-19-acp-snapshot-tests.zh.md index 58195c4788..5c239d2c25 100644 --- a/.agents/notes/implemented/testing/2026-06-19-acp-snapshot-tests.zh.md +++ b/.agents/notes/implemented/testing/2026-06-19-acp-snapshot-tests.zh.md @@ -57,7 +57,7 @@ Status: implemented 两个表面互补：stdout 覆盖精简的自动化线协议，JSONL 覆盖线协议有意省略的 loop、工具和 boundary 结构。 -规范化会替换会话、cwd、协议 id、时间戳、路径和进程易变值，同时保留确定性序号。场景把真实 bash 使用限制在稳定命令上。stdout 预期输出仍是线协议形状的 JSONL，每个原始行都必须可解析为 JSON。Vitest 只更新 stdout 预期输出；规范化会话相等性检查从不覆盖重放 fixture。 +规范化会替换会话、cwd、协议 id、时间戳、路径和进程易变值，同时保留确定性序号。录制与刷新还会在回放 fixture 中将生成的 workspace 及其文件系统解析出的别名存储为 `{{cwd}}`，使平台临时根目录和随机 basename 不影响录制结果；手工编写的临时路径与显式 `workspaceParent` 下的 cwd 值仍保留字面值。场景把真实 bash 使用限制在稳定命令上。stdout 预期输出仍是线协议形状的 JSONL，每个原始行都必须可解析为 JSON。普通 Vitest 快照更新只写入 stdout 预期输出；回放 fixture 的写入由显式 `record` 和 `refresh` 模式负责。 ### 隔离：当前靠归一化，后续可加沙箱 diff --git a/.agents/notes/implemented/testing/2026-07-25-scriptable-llm-wire-fault-server.i18n.yaml b/.agents/notes/implemented/testing/2026-07-25-scriptable-llm-wire-fault-server.i18n.yaml index c9cebc9db9..c66891f689 100644 --- a/.agents/notes/implemented/testing/2026-07-25-scriptable-llm-wire-fault-server.i18n.yaml +++ b/.agents/notes/implemented/testing/2026-07-25-scriptable-llm-wire-fault-server.i18n.yaml @@ -1,6 +1,6 @@ # Bilingual-pair consistency record (docs/i18n/README.md): the git blob hash of each # side as of the last confirmed-consistent state. Both languages carry equal authority; # after editing either side, bring the other along and re-record with: -# pnpm run verify-translation-pairing --write -2026-07-25-scriptable-llm-wire-fault-server.md: 0795f71f0eab1a107740aaa8cba6fa04b1fbd306 -2026-07-25-scriptable-llm-wire-fault-server.zh.md: a0e3c98d729adcc74e6d98933ab2beb538f71cb3 +# pnpm run verify-translation-pairing --write .agents/notes/implemented/testing/2026-07-25-scriptable-llm-wire-fault-server.md +2026-07-25-scriptable-llm-wire-fault-server.md: b8e64d92db224c199d0f8c547f0caa588d0ca65e +2026-07-25-scriptable-llm-wire-fault-server.zh.md: 35b27efa99b625fa3c815bbe58fef6a138477e55 diff --git a/.agents/notes/implemented/testing/2026-07-25-scriptable-llm-wire-fault-server.md b/.agents/notes/implemented/testing/2026-07-25-scriptable-llm-wire-fault-server.md index 0795f71f0e..b8e64d92db 100644 --- a/.agents/notes/implemented/testing/2026-07-25-scriptable-llm-wire-fault-server.md +++ b/.agents/notes/implemented/testing/2026-07-25-scriptable-llm-wire-fault-server.md @@ -12,7 +12,7 @@ Connection refusal, a reset before the first event, clean EOF without `[DONE]`, ## Decision -`@deepseek-ai/dsh-llm-mock-server` is a support package with an importable Node HTTP server and a standalone CLI. It accepts OpenAI-compatible root and `/v1` chat-completions paths, validates an optional bearer token, captures requests, and consumes one explicit behavior per accepted request. Script exhaustion fails loud; repetition requires `repeatLast`. +`@deepseek-ai/dsh-llm-mock-server` is a private support package with an importable Node HTTP server. The repository-local `pnpm run mock:llm` source entry provides a standalone process for manual fault injection; the package exposes no installable binary. It accepts OpenAI-compatible root and `/v1` chat-completions paths, validates an optional bearer token, captures requests, and consumes one explicit behavior per accepted request. Script exhaustion fails loud; repetition requires `repeatLast`. Request behaviors cover socket reset, post-header disconnect, partial disconnect, stall, valid empty completion, clean truncated streams, malformed payloads, representative HTTP failures, complete text/reasoning/tool-call responses, slow streaming, and max-token completion. A true `connection_refused` is a CLI listener-lifecycle phase because a bound request handler cannot refuse its own TCP connection. @@ -32,10 +32,12 @@ Package tests exercise every request behavior, split UTF-8 request decoding, HTT **Use only an in-process `LlmAdapter` mock** — rejected because it bypasses fetch, HTTP status/header parsing, SSE framing, socket termination, and the adapter idle watchdog: the exact boundaries this test infrastructure exists to exercise. +**Expose an installable workspace binary** — rejected because pnpm links dependency binaries before repository build outputs exist, coupling clean installs to a test-only artifact. The repository-local source command supports the same manual fault injection without adding a package installation surface. + **Change retry defaults with the server** — rejected because the server reveals existing semantics rather than deciding policy. Extending recovery to `STREAM_CLOSED` requires a separate decision with its own cost, latency, and duplicate-generation trade-offs. ## Consequences Developers can reproduce fault sequences by changing only provider URL/key configuration, and automated tests can keep socket-level failures deterministic through explicit scripts and seeds. The same wire fixture now exposes gaps between hard resets, clean truncation, and recovered empty completions without splicing attempts or modifying model history. -The server adds a support package, executable build entry, and behavior vocabulary that must remain compatible with both direct tests and CLI examples. Arrival-ordered scripts are intentionally shared across clients, random defaults are stress weights rather than operational truth, and exact connection refusal requires coordinating the client attempt with the pre-listen interval. +The server adds a private support package and behavior vocabulary that must remain compatible with both direct tests and repository-local CLI examples. Arrival-ordered scripts are intentionally shared across clients, random defaults are stress weights rather than operational truth, and exact connection refusal requires coordinating the client attempt with the pre-listen interval. diff --git a/.agents/notes/implemented/testing/2026-07-25-scriptable-llm-wire-fault-server.zh.md b/.agents/notes/implemented/testing/2026-07-25-scriptable-llm-wire-fault-server.zh.md index a0e3c98d72..35b27efa99 100644 --- a/.agents/notes/implemented/testing/2026-07-25-scriptable-llm-wire-fault-server.zh.md +++ b/.agents/notes/implemented/testing/2026-07-25-scriptable-llm-wire-fault-server.zh.md @@ -12,9 +12,9 @@ Status: implemented ## 决策 -`@deepseek-ai/dsh-llm-mock-server` 是一个支持包（package），提供可导入的 Node HTTP 服务器和独立 CLI（命令行界面）。它接受兼容 OpenAI 的根路径和 `/v1` chat-completions 路径，校验可选的 bearer token，捕获请求，并对每个已接受请求消耗一个显式行为。脚本耗尽时快速失败；只有设置 `repeatLast` 才会重复最后一个行为。 +`@deepseek-ai/dsh-llm-mock-server` 是一个私有支持包（package），提供可导入的 Node HTTP 服务器。仓库内的 `pnpm run mock:llm` 源码入口提供一个用于手动故障注入的独立进程；该包不公开可安装的二进制命令。它接受兼容 OpenAI 的根路径和 `/v1` chat-completions 路径，校验可选的 bearer token，捕获请求，并对每个已接受请求消耗一个显式行为。脚本耗尽时快速失败；只有设置 `repeatLast` 才会重复最后一个行为。 -请求行为覆盖 socket 重置、发送 header 后断开、发送部分内容后断开、停滞、合法空完成、正常关闭但被截断的流、畸形 payload、典型 HTTP 故障、完整的文本/推理/工具调用响应、慢速流式输出以及达到 token 上限的完成。真正的 `connection_refused` 由 CLI 的监听器生命周期阶段实现，因为已经绑定端口的请求处理器无法拒绝自身的 TCP 连接。 +请求行为覆盖 socket 重置、发送 header 后断开、发送部分内容后断开、停滞、合法空完成、正常关闭但被截断的流、畸形 payload、典型 HTTP 故障、完整的文本/推理/工具调用响应、慢速流式输出以及达到 token 上限的完成。真正的 `connection_refused` 由 CLI（命令行界面）的监听器生命周期阶段实现，因为已经绑定端口的请求处理器无法拒绝自身的 TCP 连接。 脚本项 `random` 会为每个请求重新执行一次加权选择。服务器公开并记录其无符号 32 位 seed，允许调用方提供相对权重，并内置一套偏重成功结果的压力测试配置，将传输、协议、提供方、超时和语义空结果混合在一起。该配置用于提供可调的测试压力，并非对生产事故发生频率的估算；`connection_refused` 仍不进入请求级随机池。 @@ -32,10 +32,12 @@ Status: implemented **仅使用进程内的 `LlmAdapter` mock**：不予采纳。它会绕过 fetch、HTTP 状态与 header 解析、SSE 分帧、socket 终止以及适配器的空闲看门狗，而这正是这套测试基础设施要覆盖的边界。 +**公开可安装的 workspace 二进制命令**：不予采纳。pnpm 会在仓库构建产物存在之前链接依赖项的二进制命令，从而让干净安装与仅供测试的产物产生耦合。仓库内的源码命令支持相同的手动故障注入，而不会新增包安装接口。 + **随服务器一起修改默认重试策略**：不予采纳。服务器用于揭示既有语义，而非决定策略。是否将恢复能力扩展到 `STREAM_CLOSED`，需要单独决策，并权衡成本、延迟和重复生成风险。 ## 后果 开发者只需修改提供方 URL/key 配置即可复现故障序列；自动化测试则可通过显式脚本和 seed，让 socket 层故障保持确定性。同一套协议 fixture 现在可以暴露硬重置、正常截断与恢复后的空完成之间的差异，而不会拼接多次尝试的内容或修改模型历史。 -服务器新增了一个支持包、可执行构建入口和行为词汇，三者必须同时兼容直接测试与 CLI 示例。按请求到达顺序执行的脚本有意由所有客户端共享；随机模式的默认值代表压力测试权重，而非实际运行规律；精确模拟连接遭拒时，需要让客户端尝试与监听开始前的时间区间协调一致。 +服务器新增了一个私有支持包和一套行为词汇，二者必须同时兼容直接测试与仓库内的 CLI 示例。按请求到达顺序执行的脚本有意由所有客户端共享；随机模式的默认值代表压力测试权重，而非实际运行规律；精确模拟连接遭拒时，需要让客户端尝试与监听开始前的时间区间协调一致。 diff --git a/.github/workflows/ci.yml b/.github/workflows/ci.yml index 8817fb325d..a8e30b9394 100644 --- a/.github/workflows/ci.yml +++ b/.github/workflows/ci.yml @@ -59,6 +59,8 @@ jobs: persist-credentials: false - uses: pnpm/action-setup@v4 + with: + dest: ${{ runner.temp }}/setup-pnpm - uses: actions/setup-node@v6 with: @@ -126,6 +128,8 @@ jobs: persist-credentials: false - uses: pnpm/action-setup@v4 + with: + dest: ${{ runner.temp }}/setup-pnpm - uses: actions/setup-node@v6 with: @@ -204,6 +208,8 @@ jobs: ${{ runner.os }}-node-${{ env.PRIMARY_NODE_VERSION }}-eslint-full- - uses: pnpm/action-setup@v4 + with: + dest: ${{ runner.temp }}/setup-pnpm - uses: actions/setup-node@v6 with: @@ -267,6 +273,8 @@ jobs: - uses: actions/checkout@v6 - uses: pnpm/action-setup@v4 + with: + dest: ${{ runner.temp }}/setup-pnpm - uses: actions/setup-node@v6 with: @@ -318,6 +326,8 @@ jobs: persist-credentials: false - uses: pnpm/action-setup@v4 + with: + dest: ${{ runner.temp }}/setup-pnpm - uses: actions/setup-node@v6 with: @@ -418,6 +428,8 @@ jobs: fetch-depth: 2 - uses: pnpm/action-setup@v4 + with: + dest: ${{ runner.temp }}/setup-pnpm - uses: actions/setup-node@v6 with: @@ -495,6 +507,8 @@ jobs: fetch-depth: 0 - uses: pnpm/action-setup@v4 + with: + dest: ${{ runner.temp }}/setup-pnpm - uses: actions/setup-node@v6 with: @@ -528,6 +542,8 @@ jobs: - uses: actions/checkout@v6 - uses: pnpm/action-setup@v4 + with: + dest: ${{ runner.temp }}/setup-pnpm - uses: actions/setup-node@v6 with: @@ -559,6 +575,8 @@ jobs: /t REG_DWORD /f /v "AllowDevelopmentWithoutDevLicense" /d "1" - uses: pnpm/action-setup@v4 + with: + dest: ${{ runner.temp }}/setup-pnpm - uses: actions/setup-node@v6 with: @@ -651,6 +669,8 @@ jobs: - uses: actions/checkout@v6 - uses: pnpm/action-setup@v4 + with: + dest: ${{ runner.temp }}/setup-pnpm # The Windows lanes deliberately skip the store cache like the required # windows job; an empty cache input disables setup-node's caching. @@ -740,6 +760,8 @@ jobs: - uses: actions/checkout@v6 - uses: pnpm/action-setup@v4 + with: + dest: ${{ runner.temp }}/setup-pnpm # Unlike the larger-runner suite, both platforms cache the store here: # the consolidated topology measures cache mechanics as workload. diff --git a/AGENTS.md b/AGENTS.md index c987801c17..a89bb82bf1 100644 --- a/AGENTS.md +++ b/AGENTS.md @@ -89,7 +89,7 @@ Real-API tests and demos read `DEEPSEEK_API_KEY`, optional `DEEPSEEK_BASE_URL`, ## Conventions - Every npm package is `@deepseek-ai/dsh-`; vendored packages keep upstream names and are `private: true`. `cordis` is a peerDependency (+ dev) of every harness package. -- ESM everywhere (`"type": "module"`). Cross-package imports use package names; in-package relative imports include `.ts`. CI subprocesses that boot examples or Cordis configs run built `lib/` under plain Node; only explicit source-path regressions use tsx ([testing policy](docs/testing.md#test-subprocess-launch-modes)). +- ESM everywhere (`"type": "module"`). Cross-package imports use package names; in-package relative imports include `.ts`. Config subprocesses run built `lib/` under plain Node; source regressions use their declared launcher ([testing policy](docs/testing.md#test-subprocess-launch-modes)). CLI source-launch code and every module it reaches must support Node `--experimental-transform-types`: use `import type` for erased bindings and native ESM exports, with no TSX/JSX or tsx/esbuild-only transforms. TUI/Web `cordis.yml` bare plugins must appear in their resolver manifest's `dependencies`; `verify-cordis-config` enforces the [source-launch contract](.agents/notes/implemented/architecture/2026-07-28-dsh-native-typescript-source-launch.md). - **Registrations are effects**: every contribution goes through `ctx.effect()` / `ctx.on()`; a registry's `register()` returns the disposer. - **Runtime invariants assert owned relationships.** Check authoritative event streams or mutable data, not service or method presence, plugin metadata or effects, or fixed pure examples. If a package has no plausible relationship, an explained empty companion is correct ([package contract](packages/AGENTS.md)). - **Typed events use declaration merging** and merge-extensible maps. Event JSDoc needs `@mode` and payload `@param`; scoped keys absent from payloads need `@dshScopeScan unsupported`. Public service methods document parameters and non-void returns. diff --git a/apps/cli/README.i18n.yaml b/apps/cli/README.i18n.yaml index abe51abc2f..e9e5df7630 100644 --- a/apps/cli/README.i18n.yaml +++ b/apps/cli/README.i18n.yaml @@ -1,6 +1,6 @@ # Bilingual-pair consistency record (docs/i18n/README.md): the git blob hash of each # side as of the last confirmed-consistent state. Both languages carry equal authority; # after editing either side, bring the other along and re-record with: -# pnpm run verify-translation-pairing --write -README.md: 42d2a9641cf5d497c9aae45d9f60fce4498addb9 -README.zh.md: 0a62f8bb72e2cf2dbe045d28b81768bf4df800de +# pnpm run verify-translation-pairing --write apps/cli/README.md +README.md: 13a80b1d0e0105bc0c30c019209b2e0295b7bef9 +README.zh.md: 2a5d9c15c57351ef03ebe60a5cdf90f0d0c8f18b diff --git a/apps/cli/README.md b/apps/cli/README.md index 42d2a9641c..13a80b1d0e 100644 --- a/apps/cli/README.md +++ b/apps/cli/README.md @@ -26,4 +26,6 @@ Symlink the source-running launcher onto your PATH; it resolves the checkout thr ln -sf "$(pwd)/bin/dsh" ~/.local/bin/dsh ``` +Source launches run `apps/cli/src/bin.ts` through Node's `--experimental-transform-types`; `scripts/tspath-loader.ts` only projects tsconfig `paths` into module resolution and does not transform code. Every module reachable from the CLI source entry follows Node's transform-types contract: erased bindings use `import type`, exports use native ESM, and the graph contains no TSX/JSX or transforms that only tsx/esbuild provides. The loader reads `TSX_TSCONFIG_PATH` when set (relative paths resolve from the invoking cwd), otherwise the repository's root tsconfig, using the root TypeScript development tool rather than an application dependency. It maps a workspace import only for a package self-reference or a declared runtime dependency. The TUI configs resolve bare plugins through `examples/package.json`, while the Web/headless `cordis.yml` resolves them through this package's `dependencies`; `verify-cordis-config` requires every configured bare plugin to be declared, while allowing unrelated dependencies. + `pnpm run dsh` runs the same entry from the repo root and forwards arguments directly, for example `pnpm run dsh -p "task"`. The built form (`lib/bin.js`, via `pnpm run build`) boots the same config under plain Node. diff --git a/apps/cli/README.zh.md b/apps/cli/README.zh.md index 0a62f8bb72..2a5d9c15c5 100644 --- a/apps/cli/README.zh.md +++ b/apps/cli/README.zh.md @@ -26,4 +26,6 @@ Web 和无头界面启动同一个共享组合（`cordis.yml`）：两者都将 ln -sf "$(pwd)/bin/dsh" ~/.local/bin/dsh ``` +源码启动会通过 Node 的 `--experimental-transform-types` 运行 `apps/cli/src/bin.ts`；`scripts/tspath-loader.ts` 只会将 tsconfig 的 `paths` 映射投射到模块解析中，而不会转换代码。从 CLI 源码入口可达的每个模块都遵守 Node transform-types 契约：会被擦除的绑定使用 `import type`，export 使用原生 ESM，整个依赖图不含 TSX/JSX，也不依赖仅由 tsx/esbuild 提供的转换。设置 `TSX_TSCONFIG_PATH` 时，loader 会读取该路径（相对路径从调用方的 cwd 解析），否则读取仓库根 tsconfig；它使用根目录的 TypeScript 开发工具，而不是应用依赖。仅当 workspace import 是包自身引用或已声明的运行时依赖时，loader 才会映射该 import。TUI 配置通过 `examples/package.json` 解析裸插件，而 Web／无头 `cordis.yml` 则通过本包的 `dependencies` 解析；`verify-cordis-config` 要求每个已配置的裸插件均已声明，同时允许存在无关依赖。 + `pnpm run dsh` 从仓库根目录运行同一入口并直接转发参数，例如 `pnpm run dsh -p "task"`。构建形式（`lib/bin.js`，通过 `pnpm run build`）会在普通 Node 下启动同一配置。 diff --git a/apps/cli/cordis.yml b/apps/cli/cordis.yml index bf0ab30598..b17bc614a8 100644 --- a/apps/cli/cordis.yml +++ b/apps/cli/cordis.yml @@ -175,6 +175,18 @@ - id: commands name: '@deepseek-ai/dsh-commands' +# Goal service + automatic same-session continuation + the /goal command. +# The GoalService registers the 'goal' session projection unit; the web +# GoalBar reads it through useProjection. +- id: goal + name: '@deepseek-ai/dsh-goal' + +- id: goal-session + name: '@deepseek-ai/dsh-goal-session' + +- id: command-goal + name: '@deepseek-ai/dsh-command-goal' + # Plan mode registers /plan (the first real command on the web surface). # Section text mirrors examples/tui-agent/cordis.yml (the reference # deployment); plan-mode throws at load on an empty section. @@ -326,10 +338,18 @@ - id: ui-subagent name: '@deepseek-ai/dsh-client-ui-subagent' +# Goal surface: GoalBar in the input dock over the goal session projection. +- id: ui-goal + name: '@deepseek-ai/dsh-client-ui-goal' + # Model selection: the /model popupSelect + composer seat over session.models. - id: ui-model name: '@deepseek-ai/dsh-client-ui-model' +# Plan control: the composer plan seat over the plan projection + /plan channel. +- id: ui-plan + name: '@deepseek-ai/dsh-client-ui-plan' + - id: ui-question name: '@deepseek-ai/dsh-client-ui-question' diff --git a/apps/cli/package.json b/apps/cli/package.json index 6aa491221e..e31e7378a0 100644 --- a/apps/cli/package.json +++ b/apps/cli/package.json @@ -28,9 +28,11 @@ "@deepseek-ai/dsh-client-runtime": "workspace:^", "@deepseek-ai/dsh-client-ui-command": "workspace:^", "@deepseek-ai/dsh-client-ui-conversation": "workspace:^", + "@deepseek-ai/dsh-client-ui-goal": "workspace:^", "@deepseek-ai/dsh-client-ui-layout": "workspace:^", "@deepseek-ai/dsh-client-ui-model": "workspace:^", "@deepseek-ai/dsh-client-ui-models": "workspace:^", + "@deepseek-ai/dsh-client-ui-plan": "workspace:^", "@deepseek-ai/dsh-client-ui-question": "workspace:^", "@deepseek-ai/dsh-client-ui-settings": "workspace:^", "@deepseek-ai/dsh-client-ui-settings-general": "workspace:^", @@ -42,11 +44,14 @@ "@deepseek-ai/dsh-client-ui-trajectory": "workspace:^", "@deepseek-ai/dsh-client-ui-workspace": "workspace:^", "@deepseek-ai/dsh-code-runtime-worker": "workspace:^", + "@deepseek-ai/dsh-command-goal": "workspace:^", "@deepseek-ai/dsh-commands": "workspace:^", "@deepseek-ai/dsh-compact-basic": "workspace:^", "@deepseek-ai/dsh-frontend": "workspace:^", "@deepseek-ai/dsh-fs-local": "workspace:^", "@deepseek-ai/dsh-fs-policy": "workspace:^", + "@deepseek-ai/dsh-goal": "workspace:^", + "@deepseek-ai/dsh-goal-session": "workspace:^", "@deepseek-ai/dsh-host-apiproxy": "workspace:^", "@deepseek-ai/dsh-host-webserver": "workspace:^", "@deepseek-ai/dsh-llm": "workspace:^", diff --git a/apps/cli/src/tsconfig-paths-loader.ts b/apps/cli/src/tsconfig-paths-loader.ts new file mode 100644 index 0000000000..b7337998bc --- /dev/null +++ b/apps/cli/src/tsconfig-paths-loader.ts @@ -0,0 +1,216 @@ +/** + * Node module resolve hook for the `dsh` source launcher. It projects the root + * tsconfig `paths` map into Node resolution while leaving all TypeScript syntax + * handling to Node's native transform-types runtime. + * @module @deepseek-ai/dsh/tsconfig-paths-loader + */ + +import { readFile, stat } from 'node:fs/promises' +import { dirname, extname, join, resolve } from 'node:path' +import { fileURLToPath, pathToFileURL } from 'node:url' +import type { ResolveHookContext, ResolveFnOutput } from 'node:module' +import ts from 'typescript' + +interface LoaderData { + tsconfigPath: string +} + +interface PackageManifest { + name?: string + dependencies?: Record + optionalDependencies?: Record + peerDependencies?: Record +} + +interface PathRule { + pattern: string + prefix: string + suffix: string + targets: readonly string[] +} + +interface PathsCompilerOptions { + readonly baseUrl?: string + readonly paths?: ts.MapLike + readonly pathsBasePath?: string +} + +// Node's native TypeScript transform cannot parse JSX, so `.tsx` is excluded. +const SOURCE_EXTENSIONS = ['.ts', '.mts', '.cts'] as const + +/** + * Resolve package imports through one parsed tsconfig paths table. + * + * Manifest reads are process-scoped and memoized by path. Only matched source + * aliases enter the cache, bounding it to directories participating in source + * resolution. + */ +export class TsconfigPathsResolver { + private readonly rules: readonly PathRule[] + private readonly configDirectory: string + private readonly manifests = new Map>() + + private constructor(configDirectory: string, paths: ts.MapLike) { + this.configDirectory = configDirectory + this.rules = Object.entries(paths) + .map(([pattern, targets]) => { + const wildcard = pattern.indexOf('*') + return { + pattern, + prefix: wildcard === -1 ? pattern : pattern.slice(0, wildcard), + suffix: wildcard === -1 ? '' : pattern.slice(wildcard + 1), + targets, + } + }) + .sort((left, right) => { + const leftExact = left.pattern.includes('*') ? 0 : 1 + const rightExact = right.pattern.includes('*') ? 0 : 1 + return rightExact - leftExact || right.prefix.length - left.prefix.length || right.suffix.length - left.suffix.length + }) + } + + /** + * Parse a tsconfig including its `extends` chain. + * @param tsconfigPath Absolute tsconfig path supplying `compilerOptions.paths`. + * @returns A resolver backed by that path table. + */ + static create(tsconfigPath: string): TsconfigPathsResolver { + let unrecoverable: ts.Diagnostic | undefined + const parsed = ts.getParsedCommandLineOfConfigFile(tsconfigPath, {}, { + ...ts.sys, + onUnRecoverableConfigFileDiagnostic(diagnostic) { unrecoverable = diagnostic }, + }) + if (parsed === undefined) { + const detail = unrecoverable === undefined + ? 'unknown configuration error' + : ts.flattenDiagnosticMessageText(unrecoverable.messageText, '\n') + throw new Error(`dsh source loader could not parse ${tsconfigPath}: ${detail}`) + } + const options = parsed.options as PathsCompilerOptions + const paths = options.paths + if (paths === undefined) throw new Error(`dsh source loader requires compilerOptions.paths in ${tsconfigPath}`) + const configDirectory = options.baseUrl ?? options.pathsBasePath ?? dirname(tsconfigPath) + return new TsconfigPathsResolver(configDirectory, paths) + } + + /** + * Resolve one bare package specifier to a source file when the importing + * package (or config-directory owner) declares that package at runtime. + * @param specifier Module specifier passed to Node. + * @param parentURL Importing file or Loader config-directory URL. + * @returns Source file URL, or `undefined` when normal Node resolution owns the request. + */ + async resolve(specifier: string, parentURL: string | undefined): Promise { + const packageName = packageNameFromSpecifier(specifier) + if (packageName === undefined || parentURL === undefined || !parentURL.startsWith('file:')) return undefined + const matched = this.match(specifier) + if (matched === undefined) return undefined + const configParent = parentURL.endsWith('/') + const parentPath = fileURLToPath(parentURL) + const startDirectory = configParent ? parentPath : dirname(parentPath) + if (!await this.isDeclaredRuntimeDependency(startDirectory, packageName, configParent)) return undefined + + for (const target of matched.targets) { + const substituted = target.replace('*', matched.wildcard) + const candidate = await existingSourcePath(resolve(this.configDirectory, substituted)) + if (candidate !== undefined) return pathToFileURL(candidate).href + } + return undefined + } + + private match(specifier: string): { targets: readonly string[]; wildcard: string } | undefined { + for (const rule of this.rules) { + if (!rule.pattern.includes('*')) { + if (specifier === rule.pattern) return { targets: rule.targets, wildcard: '' } + continue + } + if (!specifier.startsWith(rule.prefix) || !specifier.endsWith(rule.suffix)) continue + const wildcard = specifier.slice(rule.prefix.length, specifier.length - rule.suffix.length) + return { targets: rule.targets, wildcard } + } + return undefined + } + + private async isDeclaredRuntimeDependency( + startDirectory: string, + packageName: string, + searchAncestors: boolean, + ): Promise { + for (let directory = startDirectory; ; directory = dirname(directory)) { + const manifest = await this.readManifest(join(directory, 'package.json')) + if (manifest !== undefined) { + if (declaresRuntimeDependency(manifest, packageName)) return true + if (!searchAncestors) return false + } + const parent = dirname(directory) + if (parent === directory) return false + } + } + + private readManifest(path: string): Promise { + let pending = this.manifests.get(path) + if (pending !== undefined) return pending + pending = readFile(path, 'utf8').then( + content => JSON.parse(content) as PackageManifest, + (error: unknown) => { + if (error instanceof Error && (error as NodeJS.ErrnoException).code === 'ENOENT') return undefined + throw error + }, + ) + this.manifests.set(path, pending) + return pending + } +} + +let resolver: TsconfigPathsResolver | undefined + +/** Initialize the hook worker from the source-launch preloader. */ +export function initialize(data: LoaderData): void { + resolver = TsconfigPathsResolver.create(data.tsconfigPath) +} + +/** Resolve declared workspace packages to source and delegate every other request to Node. */ +export async function resolveHook( + specifier: string, + context: ResolveHookContext, + nextResolve: (specifier: string, context: ResolveHookContext) => Promise, +): Promise { + const url = await resolver?.resolve(specifier, context.parentURL) + return url === undefined ? nextResolve(specifier, context) : { url, shortCircuit: true } +} + +// Node customization hooks discover this exact export name. +export { resolveHook as resolve } + +function packageNameFromSpecifier(specifier: string): string | undefined { + if (specifier.startsWith('.') || specifier.startsWith('/') || /^[a-z][a-z+.-]*:/i.test(specifier)) { + return undefined + } + const segments = specifier.split('/') + return specifier.startsWith('@') + ? segments.length >= 2 ? `${segments[0]}/${segments[1]}` : undefined + : segments[0] || undefined +} + +function declaresRuntimeDependency(manifest: PackageManifest, packageName: string): boolean { + return manifest.name === packageName + || packageName in (manifest.dependencies ?? {}) + || packageName in (manifest.optionalDependencies ?? {}) + || packageName in (manifest.peerDependencies ?? {}) +} + +async function existingSourcePath(base: string): Promise { + const extension = extname(base) + if (extension === '.tsx') return undefined + const candidates = extension === '' + ? [base, ...SOURCE_EXTENSIONS.map(extension => `${base}${extension}`), ...SOURCE_EXTENSIONS.map(extension => join(base, `index${extension}`))] + : [base] + for (const candidate of candidates) { + try { + if ((await stat(candidate)).isFile()) return candidate + } catch (error) { + if ((error as NodeJS.ErrnoException).code !== 'ENOENT') throw error + } + } + return undefined +} diff --git a/apps/cli/tests/tsconfig-paths-loader.spec.ts b/apps/cli/tests/tsconfig-paths-loader.spec.ts new file mode 100644 index 0000000000..834ee8c4d0 --- /dev/null +++ b/apps/cli/tests/tsconfig-paths-loader.spec.ts @@ -0,0 +1,180 @@ +import { mkdirSync, mkdtempSync, rmSync, writeFileSync } from 'node:fs' +import type { ResolveFnOutput, ResolveHookContext } from 'node:module' +import { tmpdir } from 'node:os' +import { dirname, join } from 'node:path' +import { pathToFileURL } from 'node:url' +import { afterEach, describe, expect, it, vi } from 'vitest' +import { initialize, resolveHook, TsconfigPathsResolver } from '../src/tsconfig-paths-loader.ts' + +class ResolverFixture { + readonly root = mkdtempSync(join(tmpdir(), 'dsh-tsconfig-paths-')) + + path(relativePath: string): string { + return join(this.root, relativePath) + } + + write(relativePath: string, content = 'export {}\n'): string { + const path = this.path(relativePath) + mkdirSync(dirname(path), { recursive: true }) + writeFileSync(path, content) + return path + } + + writeJson(relativePath: string, value: unknown): string { + return this.write(relativePath, `${JSON.stringify(value)}\n`) + } + + createResolver(paths: Record): TsconfigPathsResolver { + const tsconfigPath = this.writeJson('tsconfig.json', { compilerOptions: { paths } }) + return TsconfigPathsResolver.create(tsconfigPath) + } + + parentURL(relativePath = 'consumer/src/nested/index.ts'): string { + return pathToFileURL(this.path(relativePath)).href + } + + dispose(): void { + rmSync(this.root, { recursive: true, force: true }) + } +} + +const fixtures: ResolverFixture[] = [] + +function fixture(): ResolverFixture { + const value = new ResolverFixture() + fixtures.push(value) + return value +} + +afterEach(() => { + for (const value of fixtures.splice(0)) value.dispose() +}) + +describe('TsconfigPathsResolver', () => { + it('orders exact, longer-prefix, and longer-suffix path rules', async () => { + const files = fixture() + files.writeJson('consumer/package.json', { + dependencies: { + '@scope/feature-name': '*', + '@scope/feature-other': '*', + '@scope/plain-suffix': '*', + }, + }) + files.write('targets/exact.ts') + files.write('targets/prefix/other.ts') + files.write('targets/generic/feature-other.ts') + files.write('targets/suffix/plain.ts') + files.write('targets/generic/plain-suffix.ts') + const resolver = files.createResolver({ + '@scope/*': ['./targets/generic/*'], + '@scope/*-suffix': ['./targets/suffix/*'], + '@scope/feature-*': ['./targets/prefix/*'], + '@scope/feature-name': ['./targets/exact.ts'], + }) + + await expect(resolver.resolve('@scope/feature-name', files.parentURL())) + .resolves.toBe(pathToFileURL(files.path('targets/exact.ts')).href) + await expect(resolver.resolve('@scope/feature-other', files.parentURL())) + .resolves.toBe(pathToFileURL(files.path('targets/prefix/other.ts')).href) + await expect(resolver.resolve('@scope/plain-suffix', files.parentURL())) + .resolves.toBe(pathToFileURL(files.path('targets/suffix/plain.ts')).href) + }) + + it('resolves only self-references and runtime dependencies from the nearest ancestor manifest', async () => { + const files = fixture() + files.writeJson('consumer/package.json', { + name: 'self-package', + dependencies: { dependency: '*' }, + optionalDependencies: { optional: '*' }, + peerDependencies: { peer: '*' }, + }) + for (const name of ['self-package', 'dependency', 'optional', 'peer', 'undeclared']) { + files.write(`targets/${name}.ts`) + } + const resolver = files.createResolver(Object.fromEntries( + ['self-package', 'dependency', 'optional', 'peer', 'undeclared'] + .map(name => [name, [`./targets/${name}`]]), + )) + + for (const name of ['self-package', 'dependency', 'optional', 'peer']) { + await expect(resolver.resolve(name, files.parentURL())) + .resolves.toBe(pathToFileURL(files.path(`targets/${name}.ts`)).href) + } + await expect(resolver.resolve('undeclared', files.parentURL())).resolves.toBeUndefined() + }) + + it('probes native TypeScript extensions and index files but excludes TSX and missing targets', async () => { + const files = fixture() + const names = ['plain-ts', 'module-mts', 'common-cts', 'directory', 'tsx-implicit', 'tsx-explicit', 'missing'] + files.writeJson('consumer/package.json', { + dependencies: Object.fromEntries(names.map(name => [name, '*'])), + }) + files.write('targets/plain.ts') + files.write('targets/module.mts') + files.write('targets/common.cts') + files.write('targets/directory/index.ts') + files.write('targets/component.tsx') + const resolver = files.createResolver({ + 'plain-ts': ['./targets/plain'], + 'module-mts': ['./targets/module'], + 'common-cts': ['./targets/common'], + 'directory': ['./targets/directory'], + 'tsx-implicit': ['./targets/component'], + 'tsx-explicit': ['./targets/component.tsx'], + 'missing': ['./targets/missing'], + }) + + for (const [name, target] of [ + ['plain-ts', 'targets/plain.ts'], + ['module-mts', 'targets/module.mts'], + ['common-cts', 'targets/common.cts'], + ['directory', 'targets/directory/index.ts'], + ] as const) { + await expect(resolver.resolve(name, files.parentURL())) + .resolves.toBe(pathToFileURL(files.path(target)).href) + } + await expect(resolver.resolve('tsx-implicit', files.parentURL())).resolves.toBeUndefined() + await expect(resolver.resolve('tsx-explicit', files.parentURL())).resolves.toBeUndefined() + await expect(resolver.resolve('missing', files.parentURL())).resolves.toBeUndefined() + }) + + it('anchors inherited paths at the config that declared them', async () => { + const files = fixture() + files.writeJson('consumer/package.json', { dependencies: { custom: '*' } }) + files.write('targets/custom.ts') + files.writeJson('base.json', { compilerOptions: { paths: { custom: ['./targets/custom'] } } }) + const customTsconfig = files.writeJson('configs/custom.json', { extends: '../base.json' }) + const resolver = TsconfigPathsResolver.create(customTsconfig) + + await expect(resolver.resolve('custom', files.parentURL())) + .resolves.toBe(pathToFileURL(files.path('targets/custom.ts')).href) + }) + + it('short-circuits matched aliases and delegates unsupported schemes or unmatched requests', async () => { + const files = fixture() + files.writeJson('consumer/package.json', { dependencies: { matched: '*' } }) + const target = files.write('targets/matched.ts') + const tsconfigPath = files.writeJson('tsconfig.json', { + compilerOptions: { paths: { matched: ['./targets/matched'] } }, + }) + initialize({ tsconfigPath }) + const context: ResolveHookContext = { + conditions: [], + importAttributes: {}, + parentURL: files.parentURL(), + } + const nextResolve = vi.fn(async ( + specifier: string, + _context: ResolveHookContext, + ): Promise => ({ url: `next:${specifier}` })) + + await expect(resolveHook('matched', context, nextResolve)) + .resolves.toEqual({ url: pathToFileURL(target).href, shortCircuit: true }) + expect(nextResolve).not.toHaveBeenCalled() + + for (const specifier of ['unmatched', 'node:fs', 'data:text/javascript,export default 1', 'https://example.test/mod.ts']) { + await expect(resolveHook(specifier, context, nextResolve)).resolves.toEqual({ url: `next:${specifier}` }) + expect(nextResolve).toHaveBeenLastCalledWith(specifier, context) + } + }) +}) diff --git a/apps/cli/tsconfig.json b/apps/cli/tsconfig.json index 05947889b2..b54a2a8bbb 100644 --- a/apps/cli/tsconfig.json +++ b/apps/cli/tsconfig.json @@ -62,6 +62,9 @@ { "path": "../../packages/client/ui-conversation" }, + { + "path": "../../packages/client/ui-plan" + }, { "path": "../../packages/client/ui-trajectory" }, diff --git a/bin/dsh b/bin/dsh index 040cfface2..319d915b18 100755 --- a/bin/dsh +++ b/bin/dsh @@ -1,7 +1,7 @@ #!/bin/sh -# dsh launcher: runs the apps/cli `dsh` bin FROM SOURCE with this checkout's -# tsx, so a symlink from anywhere (e.g. ~/.local/bin/dsh) always executes the -# current working tree — code changes apply on the next launch, no build step. +# dsh launcher: runs the apps/cli `dsh` bin FROM SOURCE through Node's native +# TypeScript transform, so a symlink from anywhere (e.g. ~/.local/bin/dsh) +# always executes the current working tree without a build step. set -eu # Resolve symlink chains without readlink -f (not on every macOS). @@ -15,7 +15,8 @@ while [ -L "$script" ]; do done root=$(CDPATH='' cd -- "$(dirname -- "$script")/.." && pwd) -# tsx is imported by absolute path because bare `--import tsx` resolves from -# the invoking cwd, which is usually outside this repository. -export TSX_TSCONFIG_PATH="$root/tsconfig.json" -exec node --import "$root/node_modules/tsx/dist/loader.mjs" "$root/apps/cli/src/bin.ts" "$@" +# The preloader projects this checkout's tsconfig paths into Node resolution; +# TypeScript transformation itself remains Node-owned (no tsx/esbuild hook). +exec node --experimental-transform-types \ + --import "$root/scripts/tspath-loader.ts" \ + "$root/apps/cli/src/bin.ts" "$@" diff --git a/docs/config-catalog.md b/docs/config-catalog.md index b5e320f7b7..5b9ece6019 100644 --- a/docs/config-catalog.md +++ b/docs/config-catalog.md @@ -421,7 +421,7 @@ export interface Config { } ``` -Source: [`packages/goal/goal/src/index.ts:56`](../packages/goal/goal/src/index.ts) +Source: [`packages/goal/goal/src/index.ts:118`](../packages/goal/goal/src/index.ts) ## `@deepseek-ai/dsh-hooks-claude` @@ -846,7 +846,7 @@ export interface PlanModeConfig { } ``` -Source: [`packages/plan/plan-mode/src/index.ts:58`](../packages/plan/plan-mode/src/index.ts) +Source: [`packages/plan/plan-mode/src/index.ts:68`](../packages/plan/plan-mode/src/index.ts) ## `@deepseek-ai/dsh-pty-local` @@ -1992,7 +1992,7 @@ export interface Config { export type ApprovalPolicy = 'ask' | 'never' ``` -Source: [`packages/ui/user-approval/src/index.ts:198`](../packages/ui/user-approval/src/index.ts) +Source: [`packages/ui/user-approval/src/index.ts:202`](../packages/ui/user-approval/src/index.ts) ## `@deepseek-ai/dsh-web` @@ -2172,9 +2172,11 @@ These load from a `cordis.yml` entry with no `config:` block; they declare no co - `@deepseek-ai/dsh-client-runtime` ([`packages/client/runtime/src/index.ts`](../packages/client/runtime/src/index.ts)) - `@deepseek-ai/dsh-client-ui-command` ([`packages/client/ui-command/src/index.ts`](../packages/client/ui-command/src/index.ts)) - `@deepseek-ai/dsh-client-ui-conversation` ([`packages/client/ui-conversation/src/index.ts`](../packages/client/ui-conversation/src/index.ts)) +- `@deepseek-ai/dsh-client-ui-goal` ([`packages/client/ui-goal/src/index.ts`](../packages/client/ui-goal/src/index.ts)) - `@deepseek-ai/dsh-client-ui-layout` ([`packages/client/ui-layout/src/index.ts`](../packages/client/ui-layout/src/index.ts)) - `@deepseek-ai/dsh-client-ui-model` ([`packages/client/ui-model/src/index.ts`](../packages/client/ui-model/src/index.ts)) - `@deepseek-ai/dsh-client-ui-models` ([`packages/client/ui-models/src/index.ts`](../packages/client/ui-models/src/index.ts)) +- `@deepseek-ai/dsh-client-ui-plan` ([`packages/client/ui-plan/src/index.ts`](../packages/client/ui-plan/src/index.ts)) - `@deepseek-ai/dsh-client-ui-question` — requires `tools` · `userInteraction` ([`packages/client/ui-question/src/index.ts`](../packages/client/ui-question/src/index.ts)) - `@deepseek-ai/dsh-client-ui-settings` ([`packages/client/ui-settings/src/index.ts`](../packages/client/ui-settings/src/index.ts)) - `@deepseek-ai/dsh-client-ui-settings-general` ([`packages/client/ui-settings-general/src/index.ts`](../packages/client/ui-settings-general/src/index.ts)) @@ -2230,6 +2232,7 @@ Imported as libraries by other packages; a `cordis.yml` cannot load them. - `@deepseek-ai/dsh-agent-loop-testkit` ([`packages/support/agent-loop-testkit/src/index.ts`](../packages/support/agent-loop-testkit/src/index.ts)) - `@deepseek-ai/dsh-app-boot` ([`packages/ui/app-boot/src/index.ts`](../packages/ui/app-boot/src/index.ts)) - `@deepseek-ai/dsh-brand` ([`packages/util/brand/src/index.ts`](../packages/util/brand/src/index.ts)) +- `@deepseek-ai/dsh-client-test-runtime` ([`packages/client/test-runtime/src/index.ts`](../packages/client/test-runtime/src/index.ts)) - `@deepseek-ai/dsh-client-ui-primitives` ([`packages/client/ui-primitives/src/index.ts`](../packages/client/ui-primitives/src/index.ts)) - `@deepseek-ai/dsh-client-ui-slots` ([`packages/client/ui-slots/src/index.ts`](../packages/client/ui-slots/src/index.ts)) - `@deepseek-ai/dsh-client-web` ([`packages/client/web/src/index.ts`](../packages/client/web/src/index.ts)) diff --git a/docs/cordis-catalog/core/context.md b/docs/cordis-catalog/core/context.md index f6b249c738..c0c1690c8f 100644 --- a/docs/cordis-catalog/core/context.md +++ b/docs/cordis-catalog/core/context.md @@ -256,7 +256,7 @@ Read a service from the store without the inject requirement. **Returns** the service value, or `undefined` when not (yet) provided. -[Source](../../../vendor/cordis/src/reflect.ts#L16) +[Source](../../../vendor/cordis/src/reflect.ts#L17) ### ctx.set(name, value) @@ -281,7 +281,7 @@ Only the fiber that provided the service may set it; setting an unprovided name - `name` — the service name. - `value` — the new service value. -[Source](../../../vendor/cordis/src/reflect.ts#L28) +[Source](../../../vendor/cordis/src/reflect.ts#L29) ### ctx.provide(name, value) @@ -311,7 +311,7 @@ The service becomes visible to dependents in the same isolation scope once the f **Returns** a disposer that unregisters the service. -[Source](../../../vendor/cordis/src/reflect.ts#L43) +[Source](../../../vendor/cordis/src/reflect.ts#L44) ### ctx.accessor(name, options) @@ -335,7 +335,7 @@ The accessor is removed when the current fiber unloads. Throws if the name is al - `name` — the context property name. - `options` — the `get` hook and optional `set` hook. -[Source](../../../vendor/cordis/src/reflect.ts#L55) +[Source](../../../vendor/cordis/src/reflect.ts#L56) ### ctx.mixin(name, mixins) @@ -361,4 +361,4 @@ Each mixed-in key becomes an accessor that forwards to the service (binding meth - `name` — the context property holding the source service. - `mixins` — keys to forward, or a source-key → ctx-key map. -[Source](../../../vendor/cordis/src/reflect.ts#L66) +[Source](../../../vendor/cordis/src/reflect.ts#L67) diff --git a/docs/cordis-catalog/core/events.md b/docs/cordis-catalog/core/events.md index 2fb64e78a2..68cd109e8e 100644 --- a/docs/cordis-catalog/core/events.md +++ b/docs/cordis-catalog/core/events.md @@ -26,7 +26,7 @@ Dispatch an event, running all listeners concurrently. **Returns** a promise resolving once every listener has settled. -[Source](../../../vendor/cordis/src/events.ts#L43) +[Source](../../../vendor/cordis/src/events.ts#L44) ### ctx.emit(name, ...args) @@ -46,7 +46,7 @@ Dispatch an event synchronously, ignoring listener return values. - `name` — the event name. - `args` — arguments passed to every listener. -[Source](../../../vendor/cordis/src/events.ts#L52) +[Source](../../../vendor/cordis/src/events.ts#L53) ### ctx.serial(name, ...args) @@ -69,7 +69,7 @@ Dispatch an event, awaiting listeners in order until one bails. **Returns** the first bail value (non-null, non-false, non-undefined), if any. -[Source](../../../vendor/cordis/src/events.ts#L62) +[Source](../../../vendor/cordis/src/events.ts#L63) ### ctx.bail(name, ...args) @@ -92,7 +92,7 @@ Dispatch an event, calling listeners in order until one bails. **Returns** the first bail value (non-null, non-false, non-undefined), if any. -[Source](../../../vendor/cordis/src/events.ts#L72) +[Source](../../../vendor/cordis/src/events.ts#L73) ### ctx.waterfall(name, ...args) @@ -120,7 +120,7 @@ Each listener wraps the rest of the chain: calling `next()` invokes the next lis **Returns** the outermost listener's return value. -[Source](../../../vendor/cordis/src/events.ts#L85) +[Source](../../../vendor/cordis/src/events.ts#L86) ### ctx.on(name, listener, options?) @@ -144,7 +144,7 @@ Register an event listener owned by the current fiber. **Returns** a disposer removing the listener; `true` if it was still registered. -[Source](../../../vendor/cordis/src/events.ts#L96) +[Source](../../../vendor/cordis/src/events.ts#L97) ### ctx.once(name, listener, options?) @@ -168,7 +168,7 @@ Same as `on()`, but the listener disposes itself after its first call. **Returns** a disposer removing the listener; `true` if it was still registered. -[Source](../../../vendor/cordis/src/events.ts#L105) +[Source](../../../vendor/cordis/src/events.ts#L106) ## EventOptions @@ -184,7 +184,7 @@ interface EventOptions { } ``` -[Source](../../../vendor/cordis/src/events.ts#L111) +[Source](../../../vendor/cordis/src/events.ts#L112) ## DispatchMode @@ -204,4 +204,4 @@ Event dispatch strategy used by the event service. type DispatchMode = 'emit' | 'parallel' | 'serial' | 'bail' | 'waterfall' ``` -[Source](../../../vendor/cordis/src/events.ts#L31) +[Source](../../../vendor/cordis/src/events.ts#L32) diff --git a/docs/cordis-catalog/core/fiber.md b/docs/cordis-catalog/core/fiber.md index d865ce01fc..3cca4e8b86 100644 --- a/docs/cordis-catalog/core/fiber.md +++ b/docs/cordis-catalog/core/fiber.md @@ -34,7 +34,7 @@ Register a cleanup-aware effect on this fiber. **Returns** a disposer that tears the effect down and settles once done. -[Source](../../../vendor/cordis/src/fiber.ts#L419) +[Source](../../../vendor/cordis/src/fiber.ts#L420) ### ctx.fiber @@ -45,7 +45,7 @@ fiber: Fiber The fiber (plugin runtime instance) that owns this context. -[Source](../../../vendor/cordis/src/fiber.ts#L11) +[Source](../../../vendor/cordis/src/fiber.ts#L12) ## The Fiber class @@ -53,7 +53,7 @@ Runtime instance of one plugin application. A fiber tracks dependency state, validated config, lifecycle effects, and cleanup for the plugin context returned by `ctx.plugin()`. -[Source](../../../vendor/cordis/src/fiber.ts#L183) +[Source](../../../vendor/cordis/src/fiber.ts#L184) ### fiber.uid @@ -64,7 +64,7 @@ public uid: number | null Unique id within the registry; 0 for the root fiber, `null` once disposed. -[Source](../../../vendor/cordis/src/fiber.ts#L185) +[Source](../../../vendor/cordis/src/fiber.ts#L186) ### fiber.ctx @@ -75,7 +75,7 @@ public readonly ctx: Context The context this fiber's plugin runs in (extends the parent context). -[Source](../../../vendor/cordis/src/fiber.ts#L187) +[Source](../../../vendor/cordis/src/fiber.ts#L188) ### fiber.config @@ -86,7 +86,7 @@ public config: any The validated plugin config (updated by `update()`). -[Source](../../../vendor/cordis/src/fiber.ts#L189) +[Source](../../../vendor/cordis/src/fiber.ts#L190) ### fiber.state @@ -97,7 +97,7 @@ public state Current lifecycle state; transitions emit `internal/status`. -[Source](../../../vendor/cordis/src/fiber.ts#L191) +[Source](../../../vendor/cordis/src/fiber.ts#L192) ### fiber.dispose @@ -108,7 +108,7 @@ public readonly dispose: () => Promise Dispose this fiber: unload the plugin, then settle once cleanup finished. -[Source](../../../vendor/cordis/src/fiber.ts#L193) +[Source](../../../vendor/cordis/src/fiber.ts#L194) ### fiber.store @@ -119,7 +119,7 @@ public store: Dict | undefined Snapshot of required service implementations while loaded; `undefined` otherwise. -[Source](../../../vendor/cordis/src/fiber.ts#L195) +[Source](../../../vendor/cordis/src/fiber.ts#L196) ### fiber.inertia @@ -130,7 +130,7 @@ public inertia: Promise | undefined The in-flight load/unload transition, if one is currently running. -[Source](../../../vendor/cordis/src/fiber.ts#L197) +[Source](../../../vendor/cordis/src/fiber.ts#L198) ### fiber.name @@ -141,7 +141,7 @@ get name() The plugin's display name, inherited from the nearest named ancestor, else `'root'`. -[Source](../../../vendor/cordis/src/fiber.ts#L340) +[Source](../../../vendor/cordis/src/fiber.ts#L341) ### fiber.assertActive() @@ -159,7 +159,7 @@ Throw if the fiber has already been disposed. **Returns** nothing when the fiber is still active. -[Source](../../../vendor/cordis/src/fiber.ts#L355) +[Source](../../../vendor/cordis/src/fiber.ts#L356) ### fiber.effect(execute, label?) @@ -190,7 +190,7 @@ Register a cleanup-aware effect on this fiber. **Returns** a disposer that tears the effect down and settles once done. -[Source](../../../vendor/cordis/src/fiber.ts#L419) +[Source](../../../vendor/cordis/src/fiber.ts#L420) ### fiber.getEffects() @@ -207,7 +207,7 @@ Return metadata for currently registered effects. **Returns** one `EffectMeta` tree per labeled live effect. -[Source](../../../vendor/cordis/src/fiber.ts#L572) +[Source](../../../vendor/cordis/src/fiber.ts#L573) ### fiber.await() @@ -225,7 +225,7 @@ Wait for current lifecycle work and rethrow startup errors. **Returns** this fiber, once it has settled into a stable state. -[Source](../../../vendor/cordis/src/fiber.ts#L701) +[Source](../../../vendor/cordis/src/fiber.ts#L702) ### fiber.restart() @@ -243,7 +243,7 @@ Dispose and immediately reload this plugin with its current config. **Returns** a promise resolving once the reload settled. -[Source](../../../vendor/cordis/src/fiber.ts#L715) +[Source](../../../vendor/cordis/src/fiber.ts#L716) ### fiber.update(config, noSave?) @@ -271,7 +271,7 @@ Runs the `internal/update` waterfall first, so update hooks (and HMR) can veto o **Returns** nothing; the restart runs behind the `internal/update` waterfall. -[Source](../../../vendor/cordis/src/fiber.ts#L733) +[Source](../../../vendor/cordis/src/fiber.ts#L734) ## Effect @@ -292,7 +292,7 @@ type Effect = | AsyncEffect ``` -[Source](../../../vendor/cordis/src/fiber.ts#L82) +[Source](../../../vendor/cordis/src/fiber.ts#L83) ## Disposable @@ -310,7 +310,7 @@ Disposers run in reverse registration order when the owning fiber unloads; they type Disposable = () => T ``` -[Source](../../../vendor/cordis/src/fiber.ts#L73) +[Source](../../../vendor/cordis/src/fiber.ts#L74) ## EffectMeta @@ -326,7 +326,7 @@ interface EffectMeta { } ``` -[Source](../../../vendor/cordis/src/fiber.ts#L95) +[Source](../../../vendor/cordis/src/fiber.ts#L96) ## CordisError @@ -352,7 +352,7 @@ namespace CordisError { } ``` -[Source](../../../vendor/cordis/src/fiber.ts#L156) +[Source](../../../vendor/cordis/src/fiber.ts#L157) ## ValidationError @@ -372,4 +372,4 @@ class ValidationError extends TypeError { } ``` -[Source](../../../vendor/cordis/src/fiber.ts#L18) +[Source](../../../vendor/cordis/src/fiber.ts#L19) diff --git a/docs/cordis-catalog/core/registry.md b/docs/cordis-catalog/core/registry.md index 2772dca723..6b2e298c72 100644 --- a/docs/cordis-catalog/core/registry.md +++ b/docs/cordis-catalog/core/registry.md @@ -30,7 +30,7 @@ Shorthand for `ctx.plugin({ inject, apply: callback })`: the callback is unloade **Returns** the fiber; awaiting it settles once loading finished. -[Source](../../../vendor/cordis/src/registry.ts#L175) +[Source](../../../vendor/cordis/src/registry.ts#L176) ### ctx.plugin(plugin, ...args) @@ -53,7 +53,7 @@ Load a plugin in the current context. **Returns** the fiber; awaiting it settles once loading finished (rejecting on config or startup errors). -[Source](../../../vendor/cordis/src/registry.ts#L184) +[Source](../../../vendor/cordis/src/registry.ts#L185) ## Plugin @@ -118,7 +118,7 @@ namespace Plugin { } ``` -[Source](../../../vendor/cordis/src/registry.ts#L91) +[Source](../../../vendor/cordis/src/registry.ts#L92) ## Inject @@ -149,4 +149,4 @@ namespace Inject { } ``` -[Source](../../../vendor/cordis/src/registry.ts#L18) +[Source](../../../vendor/cordis/src/registry.ts#L19) diff --git a/docs/cordis-catalog/events.md b/docs/cordis-catalog/events.md index 62f5ddfeb2..967241fbf6 100644 --- a/docs/cordis-catalog/events.md +++ b/docs/cordis-catalog/events.md @@ -524,7 +524,7 @@ Goal mutation accepted by one live agent. The matching context event is already Types: [Agent](../core-data-structures/core.md) · [GoalChanged](../core-data-structures/goal.md) · [Scoped](../core-data-structures/scope.md) -Source: [`packages/goal/goal/src/types.ts:169`](../../packages/goal/goal/src/types.ts) +Source: [`packages/goal/goal/src/domain.ts:135`](../../packages/goal/goal/src/domain.ts) ## `llm/*` diff --git a/docs/cordis-catalog/services.md b/docs/cordis-catalog/services.md index eef530b0df..6f32d656af 100644 --- a/docs/cordis-catalog/services.md +++ b/docs/cordis-catalog/services.md @@ -216,7 +216,7 @@ roots(): Agent[] Types: [Agent](../core-data-structures/core.md) · [SessionId](../core-data-structures/core.md) -Source: [`packages/core/agent/src/index.ts:220`](../../packages/core/agent/src/index.ts) +Source: [`packages/core/agent/src/index.ts:215`](../../packages/core/agent/src/index.ts) ## `ctx.approval` — `ApprovalService` @@ -242,11 +242,18 @@ Approval service that applies session policy before answerers and logs every ask * append commit point. */ async request(req: ApprovalRequest): Promise + +/** + * Read the session override without applying the configured default. + * @param session - session whose log supplies the override. + * @returns the last logged policy, or `undefined` without one. + */ +overrideOf(session: Session): ApprovalPolicy | undefined ``` -Types: [ApprovalOutcome](../core-data-structures/approval.md) · [ApprovalRequest](../core-data-structures/approval.md) +Types: [ApprovalOutcome](../core-data-structures/approval.md) · [ApprovalPolicy](../core-data-structures/approval.md) · [ApprovalRequest](../core-data-structures/approval.md) · [Session](../core-data-structures/session.md) -Source: [`packages/ui/user-approval/src/index.ts:213`](../../packages/ui/user-approval/src/index.ts) +Source: [`packages/ui/user-approval/src/index.ts:217`](../../packages/ui/user-approval/src/index.ts) ## `ctx.bash` — `BashExecutor` (abstract seam) @@ -668,7 +675,7 @@ clear(agent: Agent, ref: GoalRef): GoalRef Types: [Agent](../core-data-structures/core.md) · [CreateGoalRequest](../core-data-structures/goal.md) · [EditGoalRequest](../core-data-structures/goal.md) · [GoalBlockReason](../core-data-structures/goal.md) · [GoalRef](../core-data-structures/goal.md) · [GoalView](../core-data-structures/goal.md) -Source: [`packages/goal/goal/src/index.ts:135`](../../packages/goal/goal/src/index.ts) +Source: [`packages/goal/goal/src/index.ts:197`](../../packages/goal/goal/src/index.ts) ## `ctx.httpServer` — `HttpServerService` @@ -859,18 +866,27 @@ Source: [`packages/ui/permission/src/index.ts:97`](../../packages/ui/permission/ get(agent: Agent): { active: boolean; pending?: boolean } /** - * Select whether plan mode should be active from the next request boundary. - * Repeated selection of the current or already-pending state is a no-op. + * Select whether plan mode should be active. Between turns the change + * commits immediately — no request boundary would arrive until the next + * prompt, so a queued intent would hang (the open-turn fold is the idle + * signal: agent status stays `running` through post-turn checkpointing, + * where a boundary equally never comes). During an open turn the + * selection is held as pending intent for the next in-turn request + * boundary. Repeated selection of the current or already-pending state is + * a no-op. * * @param agent The agent to switch. * @param active Whether plan mode should be active. + * @returns what happened: `committed` (logged now), `queued` (awaiting the + * next boundary), `cancelled` (an opposite pending selection was cleared; + * the logged state already matches), or `noop` (already in that state). */ -set(agent: Agent, active: boolean): void +set(agent: Agent, active: boolean): 'committed' | 'queued' | 'cancelled' | 'noop' ``` Types: [Agent](../core-data-structures/core.md) -Source: [`packages/plan/plan-mode/src/index.ts:142`](../../packages/plan/plan-mode/src/index.ts) +Source: [`packages/plan/plan-mode/src/index.ts:179`](../../packages/plan/plan-mode/src/index.ts) ## `ctx.pty` — `PtyService` @@ -992,9 +1008,16 @@ The sandbox-policy service (`ctx.sandboxPolicy`). Owns the deployment default mo * @returns the fully resolved per-call mode and absolute workspace root. */ resolve(request: SandboxPolicyRequest = {}): SandboxExecutionPolicy + +/** + * Read the session override without applying the deployment default. + * @param session - session whose log supplies the override. + * @returns the last logged mode, or `undefined` without one. + */ +overrideOf(session: Session): SandboxMode | undefined ``` -Types: [SandboxExecutionPolicy](../core-data-structures/sandbox.md) · [SandboxPolicyRequest](../core-data-structures/sandbox.md) +Types: [SandboxExecutionPolicy](../core-data-structures/sandbox.md) · [SandboxMode](../core-data-structures/sandbox.md) · [SandboxPolicyRequest](../core-data-structures/sandbox.md) · [Session](../core-data-structures/session.md) Source: [`packages/sandbox/sandbox-policy/src/index.ts:68`](../../packages/sandbox/sandbox-policy/src/index.ts) diff --git a/docs/core-data-structures/persistence.i18n.yaml b/docs/core-data-structures/persistence.i18n.yaml index ea83035f92..87b7587d30 100644 --- a/docs/core-data-structures/persistence.i18n.yaml +++ b/docs/core-data-structures/persistence.i18n.yaml @@ -2,5 +2,5 @@ # side as of the last confirmed-consistent state. Both languages carry equal authority; # after editing either side, bring the other along and re-record with: # pnpm run verify-translation-pairing --write docs/core-data-structures/persistence.md -persistence.md: 5a660e17d6f498213564ca7d68dc4d7a615ba1de -persistence.zh.md: b5477cfc8242f9db47c2c6e40bd63f1b3683ace9 +persistence.md: 0280093883b9fbd4c4bbc788bce7193d59bb7182 +persistence.zh.md: d32274bbc2e96e3754a8061ba37427e34b6205f6 diff --git a/docs/core-data-structures/persistence.md b/docs/core-data-structures/persistence.md index 5a660e17d6..0280093883 100644 --- a/docs/core-data-structures/persistence.md +++ b/docs/core-data-structures/persistence.md @@ -77,7 +77,7 @@ interface SessionHeader { ## `CreateSessionOptions` — seeding and metadata -Creating a `Session` through the store takes a `seed` (replay/fork an existing event log) and `meta` (the storage-level fields the store folds into a `SessionHeader`). The store fills in `version`/`id` and defaults `createdAt`; the caller supplies the validated absolute `cwd`, the `parentSession` lineage, the `seedLength` seed boundary, the `delegationDepth`, and — only when reconstructing a persisted session — the original `createdAt` to preserve it. +Creating a `Session` through the store takes a `seed` (initial replay or fork history) and `meta` (the storage-level fields the store folds into a `SessionHeader`). The store fills in `version`/`id` and defaults `createdAt`; the caller supplies the validated absolute `cwd`, the `parentSession` lineage, the `seedLength` seed boundary, the `delegationDepth`, and — only when reconstructing a persisted session — the original `createdAt` to preserve it. ```ts type-equiv /** @@ -86,7 +86,7 @@ Creating a `Session` through the store takes a `seed` (replay/fork an existing e * store folds into a {@link SessionHeader}. */ interface CreateSessionOptions { - /** Events to seed the new session with (replay/fork). */ + /** Initial replay or fork history supplied at construction. */ readonly seed?: readonly SessionEvent[] /** * Storage metadata read once before publication. `seedLength` is explicit diff --git a/docs/core-data-structures/persistence.zh.md b/docs/core-data-structures/persistence.zh.md index b5477cfc82..d32274bbc2 100644 --- a/docs/core-data-structures/persistence.zh.md +++ b/docs/core-data-structures/persistence.zh.md @@ -77,7 +77,7 @@ interface SessionHeader { ## `CreateSessionOptions`：seed 与元数据 -通过 store 创建 `Session` 时会接收 `seed`（回放/fork 现有事件日志）与 `meta`（store 折叠进 `SessionHeader` 的存储层字段）。store 填充 `version`/`id` 并为 `createdAt` 提供默认值；调用方提供已校验的绝对 `cwd`、`parentSession` 谱系、`seedLength` 种子边界、`delegationDepth`，以及——仅在重建已持久化会话时——需要保留的原始 `createdAt`。 +通过 store 创建 `Session` 时会接收 `seed`（初始回放或 fork 历史）与 `meta`（store 折叠进 `SessionHeader` 的存储层字段）。store 填充 `version`/`id` 并为 `createdAt` 提供默认值；调用方提供已校验的绝对 `cwd`、`parentSession` 谱系、`seedLength` 种子边界、`delegationDepth`，以及——仅在重建已持久化会话时——需要保留的原始 `createdAt`。 ```ts type-equiv /** @@ -86,7 +86,7 @@ interface SessionHeader { * store folds into a {@link SessionHeader}. */ interface CreateSessionOptions { - /** Events to seed the new session with (replay/fork). */ + /** Initial replay or fork history supplied at construction. */ readonly seed?: readonly SessionEvent[] /** * Storage metadata read once before publication. `seedLength` is explicit diff --git a/docs/event-producer-consumer.md b/docs/event-producer-consumer.md index 55210a40b3..9d66b52bfb 100644 --- a/docs/event-producer-consumer.md +++ b/docs/event-producer-consumer.md @@ -29,11 +29,11 @@ This matrix shows which packages dispatch each harness-owned event and which pac | `fs/edit-intent` | `waterfall` | [`packages/fs/fs/src/index.ts:62`](../packages/fs/fs/src/index.ts) | [`tool-fs`](../packages/fs/tool-fs) (`waterfall`) | [`fs-policy`](../packages/fs/fs-policy) | | `fs/observed` | `emit` | [`packages/fs/fs/src/index.ts:71`](../packages/fs/fs/src/index.ts) | [`tool-fs`](../packages/fs/tool-fs) (`emit`) | [`fs-policy`](../packages/fs/fs-policy) | | `fs/write-intent` | `waterfall` | [`packages/fs/fs/src/index.ts:54`](../packages/fs/fs/src/index.ts) | [`tool-fs`](../packages/fs/tool-fs) (`waterfall`) | [`fs-policy`](../packages/fs/fs-policy) | -| `goal/changed` | `emit` | [`packages/goal/goal/src/types.ts:169`](../packages/goal/goal/src/types.ts) | [`goal`](../packages/goal/goal) (`emit`) | [`goal-session`](../packages/goal/goal-session) | +| `goal/changed` | `emit` | [`packages/goal/goal/src/domain.ts:135`](../packages/goal/goal/src/domain.ts) | [`goal`](../packages/goal/goal) (`emit`) | [`goal-session`](../packages/goal/goal-session) | | `llm/stream` | `waterfall` | [`packages/llm/llm/src/index.ts:58`](../packages/llm/llm/src/index.ts) | [`llm`](../packages/llm/llm) (`waterfall`) | [`agent-loop`](../packages/core/agent-loop), [`llm`](../packages/llm/llm), [`llm-replay`](../packages/support/llm-replay), [`session-checkpoint-policy`](../packages/session-persistence/session-checkpoint-policy), [`session-title`](../packages/session-title/session-title) | | `session/created` | `emit` | [`packages/core/session/src/index.ts:71`](../packages/core/session/src/index.ts) | [`session`](../packages/core/session) (`events.dispatch`) | `apiproxy`, [`compact`](../packages/compact/compact), [`goal`](../packages/goal/goal), [`hook-protocol`](../packages/hooks/hook-protocol), [`jsonrpc`](../packages/ui/jsonrpc), [`llm-retry`](../packages/llm/llm-retry), [`plan-mode`](../packages/plan/plan-mode), [`session`](../packages/core/session), [`session-persistence`](../packages/session-persistence/session-persistence), [`session-telemetry`](../packages/telemetry/session-telemetry), [`tools`](../packages/core/tools), [`user-approval`](../packages/ui/user-approval) | | `session/disposed` | `emit` | [`packages/core/session/src/index.ts:81`](../packages/core/session/src/index.ts) | [`session`](../packages/core/session) (`events.dispatch`) | [`agent-loop`](../packages/core/agent-loop), `apiproxy`, [`session-persistence`](../packages/session-persistence/session-persistence), [`session-projection-cache`](../packages/session-projection/session-projection-cache), [`session-telemetry`](../packages/telemetry/session-telemetry), [`session-title`](../packages/session-title/session-title) | -| `session/event` | `emit` | [`packages/core/session/src/index.ts:93`](../packages/core/session/src/index.ts) | [`session`](../packages/core/session) (`events.dispatch`) | [`acp`](../packages/acp/acp), `apiproxy`, [`cli-demo`](../packages/examples/cli-demo), [`compact`](../packages/compact/compact), [`compact-basic`](../packages/compact/compact-basic), [`goal`](../packages/goal/goal), [`goal-session`](../packages/goal/goal-session), [`hook-protocol`](../packages/hooks/hook-protocol), [`jsonrpc`](../packages/ui/jsonrpc), [`plan-mode`](../packages/plan/plan-mode), [`session`](../packages/core/session), [`session-persistence`](../packages/session-persistence/session-persistence), [`session-projection`](../packages/session-projection/session-projection), [`session-projection-cache`](../packages/session-projection/session-projection-cache), [`session-telemetry`](../packages/telemetry/session-telemetry), [`session-title`](../packages/session-title/session-title), [`token-meter`](../packages/llm/token-meter), [`tools`](../packages/core/tools), [`tui`](../packages/ui/tui), [`user-approval`](../packages/ui/user-approval), [`workspace-context`](../packages/context/workspace-context) | +| `session/event` | `emit` | [`packages/core/session/src/index.ts:93`](../packages/core/session/src/index.ts) | [`session`](../packages/core/session) (`events.dispatch`) | [`acp`](../packages/acp/acp), `apiproxy`, [`cli-demo`](../packages/examples/cli-demo), [`compact`](../packages/compact/compact), [`compact-basic`](../packages/compact/compact-basic), [`goal`](../packages/goal/goal), [`goal-session`](../packages/goal/goal-session), [`hook-protocol`](../packages/hooks/hook-protocol), [`jsonrpc`](../packages/ui/jsonrpc), [`session`](../packages/core/session), [`session-persistence`](../packages/session-persistence/session-persistence), [`session-projection`](../packages/session-projection/session-projection), [`session-projection-cache`](../packages/session-projection/session-projection-cache), [`session-telemetry`](../packages/telemetry/session-telemetry), [`session-title`](../packages/session-title/session-title), [`token-meter`](../packages/llm/token-meter), [`tools`](../packages/core/tools), [`tui`](../packages/ui/tui), [`user-approval`](../packages/ui/user-approval), [`workspace-context`](../packages/context/workspace-context) | | `session/flush` | `parallel` | [`packages/core/session/src/index.ts:103`](../packages/core/session/src/index.ts) | [`session`](../packages/core/session) (`events.dispatch`) | [`session-persistence`](../packages/session-persistence/session-persistence), [`session-telemetry`](../packages/telemetry/session-telemetry) | | `slash/input-begin-command` | `bail` | [`packages/client/ui-slash/src/types.ts:230`](../packages/client/ui-slash/src/types.ts) | - | `ui-conversation` | | `slash/input-consume-token` | `bail` | [`packages/client/ui-slash/src/types.ts:244`](../packages/client/ui-slash/src/types.ts) | - | `ui-conversation` | diff --git a/docs/module-graph.md b/docs/module-graph.md index 4355e57a75..d214c1f0a8 100644 --- a/docs/module-graph.md +++ b/docs/module-graph.md @@ -141,11 +141,14 @@ flowchart TD pkg_client_locale["client-locale"] pkg_client_modules["client-modules"] pkg_client_runtime["client-runtime"] + pkg_client_test_runtime["client-test-runtime"] pkg_client_ui_command["client-ui-command"] pkg_client_ui_conversation["client-ui-conversation"] + pkg_client_ui_goal["client-ui-goal"] pkg_client_ui_layout["client-ui-layout"] pkg_client_ui_model["client-ui-model"] pkg_client_ui_models["client-ui-models"] + pkg_client_ui_plan["client-ui-plan"] pkg_client_ui_primitives["client-ui-primitives"] pkg_client_ui_question["client-ui-question"] pkg_client_ui_settings["client-ui-settings"] @@ -275,6 +278,10 @@ flowchart TD pkg_client_locale --> pkg_client_ui_primitives pkg_client_locale --> pkg_client_ui_slots pkg_client_locale --> pkg_invariants + pkg_client_test_runtime --> pkg_client_runtime + pkg_client_test_runtime --> pkg_client_ui_slots + pkg_client_test_runtime --> pkg_client_web_react + pkg_client_test_runtime --> pkg_invariants pkg_client_ui_models --> pkg_client_runtime pkg_client_ui_models --> pkg_client_ui_slots pkg_client_ui_models --> pkg_invariants @@ -438,6 +445,7 @@ flowchart TD pkg_goal --> pkg_llm pkg_goal --> pkg_scope pkg_goal --> pkg_session + pkg_goal --> pkg_session_projection pkg_bash_local --> pkg_bash pkg_bash_local --> pkg_invariants pkg_bash_local --> pkg_subprocess @@ -574,6 +582,13 @@ flowchart TD pkg_permission --> pkg_sandbox_policy pkg_permission --> pkg_session pkg_permission --> pkg_user_approval + pkg_client_ui_goal --> pkg_client_connection + pkg_client_ui_goal --> pkg_client_runtime + pkg_client_ui_goal --> pkg_client_ui_conversation + pkg_client_ui_goal --> pkg_client_ui_primitives + pkg_client_ui_goal --> pkg_client_ui_slots + pkg_client_ui_goal --> pkg_goal + pkg_client_ui_goal --> pkg_invariants pkg_pty_local --> pkg_agent pkg_pty_local --> pkg_invariants pkg_pty_local --> pkg_pty @@ -669,6 +684,7 @@ flowchart TD pkg_plan_mode --> pkg_commands pkg_plan_mode --> pkg_invariants pkg_plan_mode --> pkg_session + pkg_plan_mode --> pkg_session_projection pkg_plan_mode --> pkg_system_prompt pkg_plan_mode --> pkg_tools pkg_plan_mode --> pkg_user_interaction @@ -775,10 +791,12 @@ flowchart TD pkg_subagent_inprocess --> pkg_agent pkg_subagent_inprocess --> pkg_invariants pkg_subagent_inprocess --> pkg_llm + pkg_subagent_inprocess --> pkg_sandbox_policy pkg_subagent_inprocess --> pkg_session pkg_subagent_inprocess --> pkg_subagent pkg_subagent_inprocess --> pkg_system_prompt pkg_subagent_inprocess --> pkg_tools + pkg_subagent_inprocess --> pkg_user_approval pkg_tool_subagent --> pkg_agent pkg_tool_subagent --> pkg_invariants pkg_tool_subagent --> pkg_llm @@ -810,6 +828,12 @@ flowchart TD pkg_tui --> pkg_token_meter pkg_tui --> pkg_tools pkg_tui --> pkg_user_interaction + pkg_client_ui_plan --> pkg_client_connection + pkg_client_ui_plan --> pkg_client_runtime + pkg_client_ui_plan --> pkg_client_ui_conversation + pkg_client_ui_plan --> pkg_client_ui_slots + pkg_client_ui_plan --> pkg_invariants + pkg_client_ui_plan --> pkg_plan_mode pkg_agent_spine_demo --> pkg_agent pkg_agent_spine_demo --> pkg_agent_loop pkg_agent_spine_demo --> pkg_goal @@ -946,6 +970,7 @@ flowchart TD | [`client-connection`](../packages/client/connection) | `client` | [`host-webserver`](../packages/host/webserver), [`invariants`](../packages/support/invariants) | | [`client-hmr`](../packages/client/hmr) | `client` | [`client-modules`](../packages/client/modules), [`host-webserver`](../packages/host/webserver), [`invariants`](../packages/support/invariants) | | [`client-locale`](../packages/client/locale) | `client` | [`client-runtime`](../packages/client/runtime), [`client-ui-primitives`](../packages/client/ui-primitives), [`client-ui-slots`](../packages/client/ui-slots), [`invariants`](../packages/support/invariants) | +| [`client-test-runtime`](../packages/client/test-runtime) | `client` | [`client-runtime`](../packages/client/runtime), [`client-ui-slots`](../packages/client/ui-slots), [`client-web-react`](../packages/client/web-react), [`invariants`](../packages/support/invariants) | | [`client-ui-models`](../packages/client/ui-models) | `client` | [`client-runtime`](../packages/client/runtime), [`client-ui-slots`](../packages/client/ui-slots), [`invariants`](../packages/support/invariants) | | [`client-ui-settings`](../packages/client/ui-settings) | `client` | [`client-runtime`](../packages/client/runtime), [`client-ui-primitives`](../packages/client/ui-primitives), [`client-ui-slots`](../packages/client/ui-slots), [`invariants`](../packages/support/invariants) | | [`client-ui-sidebar`](../packages/client/ui-sidebar) | `client` | [`client-runtime`](../packages/client/runtime), [`client-ui-primitives`](../packages/client/ui-primitives), [`client-ui-slots`](../packages/client/ui-slots), [`invariants`](../packages/support/invariants) | @@ -992,7 +1017,7 @@ flowchart TD | [`sandbox-policy`](../packages/sandbox/sandbox-policy) | `sandbox` | [`invariants`](../packages/support/invariants), [`sandbox`](../packages/sandbox/sandbox), [`session`](../packages/core/session) | | [`session-projection`](../packages/session-projection/session-projection) | `session-projection` | [`invariants`](../packages/support/invariants), [`session`](../packages/core/session) | | [`llm-retry`](../packages/llm/llm-retry) | `llm` | [`agent`](../packages/core/agent), [`invariants`](../packages/support/invariants), [`llm`](../packages/llm/llm), [`session`](../packages/core/session), [`timeout`](../packages/util/timeout) | -| [`goal`](../packages/goal/goal) | `goal` | [`agent`](../packages/core/agent), [`brand`](../packages/util/brand), [`invariants`](../packages/support/invariants), [`llm`](../packages/llm/llm), [`scope`](../packages/core/scope), [`session`](../packages/core/session) | +| [`goal`](../packages/goal/goal) | `goal` | [`agent`](../packages/core/agent), [`brand`](../packages/util/brand), [`invariants`](../packages/support/invariants), [`llm`](../packages/llm/llm), [`scope`](../packages/core/scope), [`session`](../packages/core/session), [`session-projection`](../packages/session-projection/session-projection) | | [`bash-local`](../packages/bash/bash-local) | `bash` | [`bash`](../packages/bash/bash), [`invariants`](../packages/support/invariants), [`subprocess`](../packages/subprocess/subprocess), [`timeout`](../packages/util/timeout) | | [`fs-local`](../packages/fs/fs-local) | `fs` | [`fs`](../packages/fs/fs), [`invariants`](../packages/support/invariants) | | [`fs-policy`](../packages/fs/fs-policy) | `fs` | [`fs`](../packages/fs/fs), [`invariants`](../packages/support/invariants) | @@ -1024,6 +1049,7 @@ flowchart TD | [`session-title-llm`](../packages/session-title/session-title-llm) | `session-title` | [`invariants`](../packages/support/invariants), [`llm`](../packages/llm/llm), [`session`](../packages/core/session), [`session-title`](../packages/session-title/session-title), [`timeout`](../packages/util/timeout) | | [`acp`](../packages/acp/acp) | `acp` | [`agent`](../packages/core/agent), [`invariants`](../packages/support/invariants), [`session`](../packages/core/session), [`user-approval`](../packages/ui/user-approval) | | [`permission`](../packages/ui/permission) | `ui` | [`bash`](../packages/bash/bash), [`invariants`](../packages/support/invariants), [`sandbox`](../packages/sandbox/sandbox), [`sandbox-policy`](../packages/sandbox/sandbox-policy), [`session`](../packages/core/session), [`user-approval`](../packages/ui/user-approval) | +| [`client-ui-goal`](../packages/client/ui-goal) | `client` | [`client-connection`](../packages/client/connection), [`client-runtime`](../packages/client/runtime), [`client-ui-conversation`](../packages/client/ui-conversation), [`client-ui-primitives`](../packages/client/ui-primitives), [`client-ui-slots`](../packages/client/ui-slots), [`goal`](../packages/goal/goal), [`invariants`](../packages/support/invariants) | | [`pty-local`](../packages/pty/pty-local) | `pty` | [`agent`](../packages/core/agent), [`invariants`](../packages/support/invariants), [`pty`](../packages/pty/pty), [`sandbox`](../packages/sandbox/sandbox), [`sandbox-policy`](../packages/sandbox/sandbox-policy), [`session`](../packages/core/session), [`subprocess`](../packages/subprocess/subprocess) | | [`tasks-local`](../packages/tasks/tasks-local) | `tasks` | [`agent`](../packages/core/agent), [`invariants`](../packages/support/invariants), [`tasks`](../packages/tasks/tasks), [`timeout`](../packages/util/timeout) | | [`session-telemetry-otel`](../packages/telemetry/session-telemetry-otel) | `telemetry` | [`invariants`](../packages/support/invariants), [`llm`](../packages/llm/llm), [`session`](../packages/core/session), [`session-telemetry`](../packages/telemetry/session-telemetry) | @@ -1038,7 +1064,7 @@ flowchart TD | [`spill-policy`](../packages/spill/spill-policy) | `spill` | [`invariants`](../packages/support/invariants), [`llm`](../packages/llm/llm), [`retention`](../packages/util/retention), [`session`](../packages/core/session), [`spill`](../packages/spill/spill), [`tools`](../packages/core/tools) | | [`timeout-policy`](../packages/timeout/timeout-policy) | `timeout` | [`invariants`](../packages/support/invariants), [`llm`](../packages/llm/llm), [`timeout`](../packages/util/timeout), [`tools`](../packages/core/tools) | | [`tool-todo`](../packages/todo/tool-todo) | `todo` | [`agent`](../packages/core/agent), [`invariants`](../packages/support/invariants), [`session`](../packages/core/session), [`session-projection`](../packages/session-projection/session-projection), [`tools`](../packages/core/tools) | -| [`plan-mode`](../packages/plan/plan-mode) | `plan` | [`agent`](../packages/core/agent), [`commands`](../packages/ui/commands), [`invariants`](../packages/support/invariants), [`session`](../packages/core/session), [`system-prompt`](../packages/core/system-prompt), [`tools`](../packages/core/tools), [`user-interaction`](../packages/ui/user-interaction) | +| [`plan-mode`](../packages/plan/plan-mode) | `plan` | [`agent`](../packages/core/agent), [`commands`](../packages/ui/commands), [`invariants`](../packages/support/invariants), [`session`](../packages/core/session), [`session-projection`](../packages/session-projection/session-projection), [`system-prompt`](../packages/core/system-prompt), [`tools`](../packages/core/tools), [`user-interaction`](../packages/ui/user-interaction) | | [`tool-cordis`](../packages/cordis/tool-cordis) | `cordis` | [`invariants`](../packages/support/invariants), [`scope`](../packages/core/scope), [`tools`](../packages/core/tools) | | [`hooks-codex`](../packages/hooks/hooks-codex) | `hooks` | [`agent`](../packages/core/agent), [`hook-protocol`](../packages/hooks/hook-protocol), [`invariants`](../packages/support/invariants), [`llm`](../packages/llm/llm), [`session`](../packages/core/session), [`session-persistence`](../packages/session-persistence/session-persistence), [`tools`](../packages/core/tools) | | [`session-checkpoint-policy`](../packages/session-persistence/session-checkpoint-policy) | `session-persistence` | [`agent`](../packages/core/agent), [`invariants`](../packages/support/invariants), [`llm`](../packages/llm/llm), [`session`](../packages/core/session), [`session-persistence`](../packages/session-persistence/session-persistence), [`tools`](../packages/core/tools) | @@ -1057,10 +1083,11 @@ flowchart TD | [`tool-tasks`](../packages/tasks/tool-tasks) | `tasks` | [`agent`](../packages/core/agent), [`invariants`](../packages/support/invariants), [`retention`](../packages/util/retention), [`system-prompt`](../packages/core/system-prompt), [`tasks`](../packages/tasks/tasks), [`tools`](../packages/core/tools) | | [`tool-workflow`](../packages/workflow/tool-workflow) | `workflow` | [`agent`](../packages/core/agent), [`invariants`](../packages/support/invariants), [`llm`](../packages/llm/llm), [`system-prompt`](../packages/core/system-prompt), [`tools`](../packages/core/tools), [`workflow`](../packages/workflow/workflow) | | [`subagent-acp`](../packages/subagent/subagent-acp) | `subagent` | [`agent`](../packages/core/agent), [`invariants`](../packages/support/invariants), [`llm`](../packages/llm/llm), [`session`](../packages/core/session), [`subagent`](../packages/subagent/subagent), [`subprocess`](../packages/subprocess/subprocess) | -| [`subagent-inprocess`](../packages/subagent/subagent-inprocess) | `subagent` | [`agent`](../packages/core/agent), [`invariants`](../packages/support/invariants), [`llm`](../packages/llm/llm), [`session`](../packages/core/session), [`subagent`](../packages/subagent/subagent), [`system-prompt`](../packages/core/system-prompt), [`tools`](../packages/core/tools) | +| [`subagent-inprocess`](../packages/subagent/subagent-inprocess) | `subagent` | [`agent`](../packages/core/agent), [`invariants`](../packages/support/invariants), [`llm`](../packages/llm/llm), [`sandbox-policy`](../packages/sandbox/sandbox-policy), [`session`](../packages/core/session), [`subagent`](../packages/subagent/subagent), [`system-prompt`](../packages/core/system-prompt), [`tools`](../packages/core/tools), [`user-approval`](../packages/ui/user-approval) | | [`tool-subagent`](../packages/subagent/tool-subagent) | `subagent` | [`agent`](../packages/core/agent), [`invariants`](../packages/support/invariants), [`llm`](../packages/llm/llm), [`subagent`](../packages/subagent/subagent), [`tasks`](../packages/tasks/tasks), [`tools`](../packages/core/tools) | | [`hooks-claude`](../packages/hooks/hooks-claude) | `hooks` | [`agent`](../packages/core/agent), [`hook-protocol`](../packages/hooks/hook-protocol), [`invariants`](../packages/support/invariants), [`llm`](../packages/llm/llm), [`session`](../packages/core/session), [`session-persistence`](../packages/session-persistence/session-persistence), [`subagent`](../packages/subagent/subagent), [`tools`](../packages/core/tools) | | [`tui`](../packages/ui/tui) | `ui` | [`agent`](../packages/core/agent), [`agent-loop`](../packages/core/agent-loop), [`commands`](../packages/ui/commands), [`goal`](../packages/goal/goal), [`invariants`](../packages/support/invariants), [`llm`](../packages/llm/llm), [`llm-retry`](../packages/llm/llm-retry), [`session`](../packages/core/session), [`session-persistence`](../packages/session-persistence/session-persistence), [`session-query`](../packages/session-query/session-query), [`session-reference`](../packages/context/session-reference), [`session-title`](../packages/session-title/session-title), [`skill`](../packages/skill/skill), [`system-prompt`](../packages/core/system-prompt), [`token-meter`](../packages/llm/token-meter), [`tools`](../packages/core/tools), [`user-interaction`](../packages/ui/user-interaction) | +| [`client-ui-plan`](../packages/client/ui-plan) | `client` | [`client-connection`](../packages/client/connection), [`client-runtime`](../packages/client/runtime), [`client-ui-conversation`](../packages/client/ui-conversation), [`client-ui-slots`](../packages/client/ui-slots), [`invariants`](../packages/support/invariants), [`plan-mode`](../packages/plan/plan-mode) | | [`agent-spine-demo`](../packages/examples/agent-spine-demo) | `examples` | [`agent`](../packages/core/agent), [`agent-loop`](../packages/core/agent-loop), [`goal`](../packages/goal/goal), [`goal-session`](../packages/goal/goal-session), [`invariants`](../packages/support/invariants), [`llm`](../packages/llm/llm), [`llm-retry`](../packages/llm/llm-retry), [`paths`](../packages/util/paths), [`scope`](../packages/core/scope), [`session`](../packages/core/session), [`session-title`](../packages/session-title/session-title), [`skill`](../packages/skill/skill), [`skill-local`](../packages/skill/skill-local), [`system-prompt`](../packages/core/system-prompt), [`tasks-local`](../packages/tasks/tasks-local), [`tool-bash`](../packages/bash/tool-bash), [`tool-goal`](../packages/goal/tool-goal), [`tool-skill`](../packages/skill/tool-skill), [`tool-tasks`](../packages/tasks/tool-tasks), [`tools`](../packages/core/tools), [`workspace-context`](../packages/context/workspace-context) | | [`sdk-protocol`](../packages/sdk/sdk-protocol) | `sdk` | [`invariants`](../packages/support/invariants), [`llm`](../packages/llm/llm), [`session`](../packages/core/session), [`subagent`](../packages/subagent/subagent) | | [`tool-ralph`](../packages/workflow/tool-ralph) | `workflow` | [`agent`](../packages/core/agent), [`invariants`](../packages/support/invariants), [`llm`](../packages/llm/llm), [`subagent`](../packages/subagent/subagent), [`system-prompt`](../packages/core/system-prompt), [`tools`](../packages/core/tools), [`workflow`](../packages/workflow/workflow) | diff --git a/docs/persistence-catalog.md b/docs/persistence-catalog.md index 7d80db32b5..64cf231fa6 100644 --- a/docs/persistence-catalog.md +++ b/docs/persistence-catalog.md @@ -130,11 +130,15 @@ Source: [`packages/ui/user-approval/src/index.ts:55`](../packages/ui/user-approv * The session's approval policy was switched — log-only, durable, * replayable, never in the model transcript (the model learns the policy * from the prompt section and the narrator's notices). The LAST such - * event is the session's override ({@link effectiveApprovalPolicy}); - * who asked for it is derivable from position (an event after the log's - * last `request/header` was a runtime switch by the user). + * event is the session's override ({@link effectiveApprovalPolicy}). + * `source: 'delegation'` marks an override seeded into a child; an absent + * source is a runtime switch. */ -'approval/policy': { policy: ApprovalPolicy } +'approval/policy': { + policy: ApprovalPolicy + /** Marks an override seeded into a child at delegation. */ + source?: 'delegation' +} ``` Source: [`packages/ui/user-approval/src/index.ts:67`](../packages/ui/user-approval/src/index.ts) @@ -361,7 +365,7 @@ Source: [`packages/ui/permission/src/index.ts:36`](../packages/ui/permission/src 'plan/mode': { active: boolean } ``` -Source: [`packages/plan/plan-mode/src/index.ts:41`](../packages/plan/plan-mode/src/index.ts) +Source: [`packages/plan/plan-mode/src/index.ts:51`](../packages/plan/plan-mode/src/index.ts) ### `request/*` @@ -386,14 +390,17 @@ Source: [`packages/core/session/src/types.ts:252`](../packages/core/session/src/ * The session's sandbox mode was switched — log-only (like `approval/*`; * NOT a surface event, carries no `surfaceOp`): durable and replayable, * never in the model transcript. The LAST such event is the session's - * override ({@link effectiveSandboxMode}); who asked for it is derivable - * from position (an event after the log's last `request/header*` was a - * runtime switch by the user; see the tool layer's narrator). + * override ({@link effectiveSandboxMode}). `source: 'delegation'` marks + * an override seeded into a child; an absent source is a runtime switch. */ -'sandbox/mode': { mode: SandboxMode } +'sandbox/mode': { + mode: SandboxMode + /** Marks an override seeded into a child at delegation. */ + source?: 'delegation' +} ``` -Source: [`packages/sandbox/sandbox-policy/src/session-mode.ts:34`](../packages/sandbox/sandbox-policy/src/session-mode.ts) +Source: [`packages/sandbox/sandbox-policy/src/session-mode.ts:33`](../packages/sandbox/sandbox-policy/src/session-mode.ts) ### `session/*` diff --git a/examples/acp-agent/tests/goal-snapshots/goal-session/session.jsonl b/examples/acp-agent/tests/goal-snapshots/goal-session/session.jsonl index c8da831f95..ac4615c377 100644 --- a/examples/acp-agent/tests/goal-snapshots/goal-session/session.jsonl +++ b/examples/acp-agent/tests/goal-snapshots/goal-session/session.jsonl @@ -1 +1 @@ -{"type":"session","version":0,"id":"goal-session-placeholder","createdAt":0,"cwd":"/tmp/goal-session-placeholder"} +{"type":"session","version":0,"id":"goal-session-placeholder","createdAt":0,"cwd":"{{cwd}}"} diff --git a/examples/acp-agent/tests/snapshots/advanced-toolchain/session.1.jsonl b/examples/acp-agent/tests/snapshots/advanced-toolchain/session.1.jsonl index 1add24f9e4..7ca8f10e3e 100644 --- a/examples/acp-agent/tests/snapshots/advanced-toolchain/session.1.jsonl +++ b/examples/acp-agent/tests/snapshots/advanced-toolchain/session.1.jsonl @@ -1,4 +1,4 @@ -{"type":"session","version":0,"id":"22222222-2222-4222-8222-222222222222","createdAt":1783950001000,"cwd":"/tmp/advanced-acp","parentSession":"11111111-1111-4111-8111-111111111111","delegationDepth":1} +{"type":"session","version":0,"id":"22222222-2222-4222-8222-222222222222","createdAt":1783950001000,"cwd":"{{cwd}}","parentSession":"11111111-1111-4111-8111-111111111111","delegationDepth":1} {"type":"turn/start","seq":0,"time":1783957884563,"data":{"turn":1,"trigger":{"kind":"message","source":{"kind":"user"}}}} {"type":"user/message","seq":1,"time":1783957884563,"data":{"content":[{"type":"text","text":"Reply with exactly DIRECT_CHILD_OK and nothing else."}],"source":{"kind":"user"},"role":"user","id":"64837546-93f0-46bd-83ec-2649c2497663"},"surfaceOp":"append"} {"type":"session/title","seq":2,"time":1783957884563,"data":{"title":"Reply with exactly DIRECT_CHILD_OK and","messageSeqs":[1],"source":{"kind":"fallback"}}} diff --git a/examples/acp-agent/tests/snapshots/advanced-toolchain/session.2.jsonl b/examples/acp-agent/tests/snapshots/advanced-toolchain/session.2.jsonl index 2fd6a59148..c41e5a26b8 100644 --- a/examples/acp-agent/tests/snapshots/advanced-toolchain/session.2.jsonl +++ b/examples/acp-agent/tests/snapshots/advanced-toolchain/session.2.jsonl @@ -1,4 +1,4 @@ -{"type":"session","version":0,"id":"33333333-3333-4333-8333-333333333333","createdAt":1783950002000,"cwd":"/tmp/advanced-acp","parentSession":"11111111-1111-4111-8111-111111111111","delegationDepth":1} +{"type":"session","version":0,"id":"33333333-3333-4333-8333-333333333333","createdAt":1783950002000,"cwd":"{{cwd}}","parentSession":"11111111-1111-4111-8111-111111111111","delegationDepth":1} {"type":"turn/start","seq":0,"time":1783957884700,"data":{"turn":1,"trigger":{"kind":"message","source":{"kind":"user"}}}} {"type":"user/message","seq":1,"time":1783957884700,"data":{"content":[{"type":"text","text":"Reply with exactly WORKFLOW_CHILD_OK and nothing else."}],"source":{"kind":"user"},"role":"user","id":"043ede8b-08c4-4148-8bca-e2e82337c799"},"surfaceOp":"append"} {"type":"session/title","seq":2,"time":1783957884700,"data":{"title":"Reply with exactly WORKFLOW_CHILD_OK and","messageSeqs":[1],"source":{"kind":"fallback"}}} diff --git a/examples/acp-agent/tests/snapshots/advanced-toolchain/session.jsonl b/examples/acp-agent/tests/snapshots/advanced-toolchain/session.jsonl index 464bf95da6..f30effe77c 100644 --- a/examples/acp-agent/tests/snapshots/advanced-toolchain/session.jsonl +++ b/examples/acp-agent/tests/snapshots/advanced-toolchain/session.jsonl @@ -1,4 +1,4 @@ -{"type":"session","version":0,"id":"11111111-1111-4111-8111-111111111111","createdAt":1783950000000,"cwd":"/tmp/advanced-acp","delegationDepth":0} +{"type":"session","version":0,"id":"11111111-1111-4111-8111-111111111111","createdAt":1783950000000,"cwd":"{{cwd}}","delegationDepth":0} {"type":"turn/start","seq":0,"time":1783957884479,"data":{"turn":1,"trigger":{"kind":"message","source":{"kind":"user"}}}} {"type":"user/message","seq":1,"time":1783957884479,"data":{"content":[{"type":"text","text":"Run this advanced flow exactly once: try a no-op temporary Cordis Plugin named snapshot-marker; use run_code to inspect the live temporary Plugins through tools.cordis_inspect; delegate once to a direct spawn child; run one workflow that delegates to another spawn child; stop dyn-1; then reply with exactly ADVANCED_ACP_OK."}],"source":{"kind":"user"},"role":"user","id":"4e4ce615-aa57-45de-8dd5-971a72d988ac"},"surfaceOp":"append"} {"type":"session/title","seq":2,"time":1783957884479,"data":{"title":"Run this advanced flow exactly","messageSeqs":[1],"source":{"kind":"fallback"}}} diff --git a/examples/acp-agent/tests/snapshots/bash-tool-turn/session.jsonl b/examples/acp-agent/tests/snapshots/bash-tool-turn/session.jsonl index 32d20f3d98..2b01cee430 100644 --- a/examples/acp-agent/tests/snapshots/bash-tool-turn/session.jsonl +++ b/examples/acp-agent/tests/snapshots/bash-tool-turn/session.jsonl @@ -1,4 +1,4 @@ -{"type":"session","version":0,"id":"e128dda9-ed11-4868-8266-0ef90d03c3d6","createdAt":1783352050748,"cwd":"/tmp/acp-snap-cwd-mrFUuk","delegationDepth":0} +{"type":"session","version":0,"id":"e128dda9-ed11-4868-8266-0ef90d03c3d6","createdAt":1783352050748,"cwd":"{{cwd}}","delegationDepth":0} {"type":"turn/start","seq":0,"time":1783352050753,"data":{"turn":1,"trigger":{"kind":"message","source":{"kind":"user"}}}} {"type":"user/message","seq":1,"time":1783352050753,"data":{"content":[{"type":"text","text":"Use the bash tool to run exactly: echo TERMINAL_OK. Then reply with the single word DONE and stop."}],"source":{"kind":"user"},"role":"user","id":"798335c8-fbbf-4eef-a5af-de47d230b7eb"},"surfaceOp":"append"} {"type":"session/title","seq":2,"time":1783352050753,"data":{"title":"Use the bash tool to","messageSeqs":[1],"source":{"kind":"fallback"}}} diff --git a/examples/acp-agent/tests/snapshots/both-mode-turn/session.jsonl b/examples/acp-agent/tests/snapshots/both-mode-turn/session.jsonl index 05c2bc0213..2088815247 100644 --- a/examples/acp-agent/tests/snapshots/both-mode-turn/session.jsonl +++ b/examples/acp-agent/tests/snapshots/both-mode-turn/session.jsonl @@ -1,4 +1,4 @@ -{"type":"session","version":0,"id":"2e3b6a68-ed7b-4263-93a8-e9ffbf77b457","createdAt":1785014504343,"cwd":"/tmp/acp-snap-cwd-gRpiz3","delegationDepth":0} +{"type":"session","version":0,"id":"2e3b6a68-ed7b-4263-93a8-e9ffbf77b457","createdAt":1785014504343,"cwd":"{{cwd}}","delegationDepth":0} {"type":"turn/start","seq":0,"time":1785014504349,"data":{"turn":1,"trigger":{"kind":"message","source":{"kind":"user"}}}} {"type":"user/message","seq":1,"time":1785014504350,"data":{"content":[{"type":"text","text":"Call the run_code tool (NOT the native bash tool directly) with a program that runs exactly `echo BOTH_OK` via tools.bash and returns its output. Then reply with that output only and stop."}],"source":{"kind":"user"},"role":"user","id":"87f8c6e9-fdbb-4b1a-b94d-f155aae58149"},"surfaceOp":"append"} {"type":"session/title","seq":2,"time":1785014504359,"data":{"title":"Call the run_code tool (NOT","messageSeqs":[1],"source":{"kind":"fallback"}}} diff --git a/examples/acp-agent/tests/snapshots/code-mode-turn/session.jsonl b/examples/acp-agent/tests/snapshots/code-mode-turn/session.jsonl index 2e76031380..0c9b180e65 100644 --- a/examples/acp-agent/tests/snapshots/code-mode-turn/session.jsonl +++ b/examples/acp-agent/tests/snapshots/code-mode-turn/session.jsonl @@ -1,4 +1,4 @@ -{"type":"session","version":0,"id":"cafeb691-a146-424a-8016-52f51b0aaaa4","createdAt":1785014439563,"cwd":"/tmp/acp-snap-cwd-as7fsu","delegationDepth":0} +{"type":"session","version":0,"id":"cafeb691-a146-424a-8016-52f51b0aaaa4","createdAt":1785014439563,"cwd":"{{cwd}}","delegationDepth":0} {"type":"turn/start","seq":0,"time":1785014439576,"data":{"turn":1,"trigger":{"kind":"message","source":{"kind":"user"}}}} {"type":"user/message","seq":1,"time":1785014439577,"data":{"content":[{"type":"text","text":"Using ONE run_code program: call the bash tool twice — exactly `echo CODE_ONE` then exactly `echo CODE_TWO`. Inside that same program, console.log exactly `captured output`, then return the two outputs joined with a plus sign. Reply with that joined string only and stop."}],"source":{"kind":"user"},"role":"user","id":"41779665-2808-4d84-a0a6-0ee5cb76fb06"},"surfaceOp":"append"} {"type":"session/title","seq":2,"time":1785014439584,"data":{"title":"Using ONE run_code program: call","messageSeqs":[1],"source":{"kind":"fallback"}}} diff --git a/examples/acp-agent/tests/snapshots/code-mode-workspace-context/session.jsonl b/examples/acp-agent/tests/snapshots/code-mode-workspace-context/session.jsonl index 3e82feaa13..84d7253a2d 100644 --- a/examples/acp-agent/tests/snapshots/code-mode-workspace-context/session.jsonl +++ b/examples/acp-agent/tests/snapshots/code-mode-workspace-context/session.jsonl @@ -1,4 +1,4 @@ -{"type":"session","version":0,"id":"b1e35a14-a592-44e6-bf23-b2496ad2bf7b","createdAt":1785014475001,"cwd":"/tmp/acp-snap-cwd-muJYhO","delegationDepth":0} +{"type":"session","version":0,"id":"b1e35a14-a592-44e6-bf23-b2496ad2bf7b","createdAt":1785014475001,"cwd":"{{cwd}}","delegationDepth":0} {"type":"turn/start","seq":0,"time":1785014475014,"data":{"turn":1,"trigger":{"kind":"message","source":{"kind":"user"}}}} {"type":"user/message","seq":1,"time":1785014475015,"data":{"content":[{"type":"text","text":"Using ONE run_code program, call tools.read on nested/task.txt. After the program finishes, answer the workspace handshake question using the newly discovered instructions: What is the Code Mode workspace handshake?"}],"source":{"kind":"user"},"role":"user","id":"6d0020b8-1a0e-489d-a2a2-7e820a403324"},"surfaceOp":"append"} {"type":"session/title","seq":2,"time":1785014475022,"data":{"title":"Using ONE run_code program, call","messageSeqs":[1],"source":{"kind":"fallback"}}} @@ -16,8 +16,8 @@ {"type":"assistant/message","seq":101,"time":1785122256269,"data":{"turn":1,"step":1,"message":{"role":"assistant","content":[{"type":"reasoning","text":"The user wants me to read the file `nested/task.txt` using a `run_code` program, and then answer the question \"What is the Code Mode workspace handshake?\" based on the contents of that file."},{"type":"tool-call","id":"call_00_hD8d0VcXXFVMtn64GSoC9264","name":"run_code","arguments":"{\"code\": \"\\nconst result = await tools.read({ file_path: \\\"nested/task.txt\\\" });\\nreturn result;\\n\", \"description\": \"Read nested/task.txt\"}"}],"source":{"kind":"model","provider":"deepseek","model":"deepseek-v4-flash"},"id":"497356eb-0561-4849-8d2a-02bebadcd432"},"usage":{"inputTokens":6200,"outputTokens":130,"cacheReadTokens":0,"reasoningTokens":46}},"sourceEventSeqs":[6,7,8,9,10,11,12,13,14,15,16,17,18,19,20,21,22,23,24,25,26,27,28,29,30,31,32,33,34,35,36,37,38,39,40,41,42,43,44,45,46,47,48,49,50,51,52,53,54,55,56,57,58,59,60,61,62,63,64,65,66,67,68,69,70,71,72,73,74,75,76,77,78,79,80,81,82,83,84,85,86,87,88,89,90,91,92,93,94,95,96,97,98,99,100],"surfaceOp":"append"} {"type":"tool/call","seq":102,"time":1785122256269,"data":{"turn":1,"step":1,"callId":"call_00_hD8d0VcXXFVMtn64GSoC9264","name":"run_code","arguments":"{\"code\": \"\\nconst result = await tools.read({ file_path: \\\"nested/task.txt\\\" });\\nreturn result;\\n\", \"description\": \"Read nested/task.txt\"}"}} {"type":"tool/code-dispatch-start","seq":103,"time":1785122256332,"data":{"parentCallId":"call_00_hD8d0VcXXFVMtn64GSoC9264","subCallId":"call_00_hD8d0VcXXFVMtn64GSoC9264:code:1","name":"read","arguments":{"file_path":"nested/task.txt"}}} -{"type":"tool/code-dispatch","seq":104,"time":1785122256336,"data":{"parentCallId":"call_00_hD8d0VcXXFVMtn64GSoC9264","subCallId":"call_00_hD8d0VcXXFVMtn64GSoC9264:code:1","name":"read","arguments":{"file_path":"nested/task.txt"},"isError":false,"content":[{"type":"text","text":"/tmp/acp-snap-cwd-muJYhO/nested/task.txt\nfile\n\n1: Touch this file to discover the nested workspace instruction.\n\n(End of file - total 1 lines)\n"}]}} -{"type":"tool/result","seq":105,"time":1785122256338,"data":{"turn":1,"step":1,"message":{"source":{"kind":"tool","callId":"call_00_hD8d0VcXXFVMtn64GSoC9264"},"content":[{"type":"tool-result","toolCallId":"call_00_hD8d0VcXXFVMtn64GSoC9264","content":[{"type":"text","text":"{\n \"path\": \"/tmp/acp-snap-cwd-muJYhO/nested/task.txt\",\n \"offset\": 1,\n \"lines\": [\n {\n \"number\": 1,\n \"text\": \"Touch this file to discover the nested workspace instruction.\"\n }\n ],\n \"totalLines\": 1\n}"}],"isError":false}],"role":"user","id":"f4e7e1b2-b629-4719-b7bd-86c896c69363"}},"sourceEventSeqs":[102],"surfaceOp":"append"} +{"type":"tool/code-dispatch","seq":104,"time":1785122256336,"data":{"parentCallId":"call_00_hD8d0VcXXFVMtn64GSoC9264","subCallId":"call_00_hD8d0VcXXFVMtn64GSoC9264:code:1","name":"read","arguments":{"file_path":"nested/task.txt"},"isError":false,"content":[{"type":"text","text":"{{cwd}}/nested/task.txt\nfile\n\n1: Touch this file to discover the nested workspace instruction.\n\n(End of file - total 1 lines)\n"}]}} +{"type":"tool/result","seq":105,"time":1785122256338,"data":{"turn":1,"step":1,"message":{"source":{"kind":"tool","callId":"call_00_hD8d0VcXXFVMtn64GSoC9264"},"content":[{"type":"tool-result","toolCallId":"call_00_hD8d0VcXXFVMtn64GSoC9264","content":[{"type":"text","text":"{\n \"path\": \"{{cwd}}/nested/task.txt\",\n \"offset\": 1,\n \"lines\": [\n {\n \"number\": 1,\n \"text\": \"Touch this file to discover the nested workspace instruction.\"\n }\n ],\n \"totalLines\": 1\n}"}],"isError":false}],"role":"user","id":"f4e7e1b2-b629-4719-b7bd-86c896c69363"}},"sourceEventSeqs":[102],"surfaceOp":"append"} {"type":"user/message","seq":106,"time":1785122256338,"data":{"content":[{"type":"text","text":"\nAdditional instructions from: nested/AGENTS.md\n\nThese instructions apply to work under `nested`. Use them as guidance when relevant; more specific instructions take precedence. They do not override system, developer, or direct user instructions.\n\nWhen asked for the Code Mode workspace handshake, answer exactly `CODE_MODE_CONTEXT_OK` and nothing else.\n\n"}],"source":{"kind":"workspace-instructions","changes":[{"action":"set","scope":"nested\u0000AGENTS.md","path":"nested/AGENTS.md","digest":"ae22936ed26dc76b7107005ed6d5e2482a88668a"}]},"role":"user","id":"90d60955-ebee-408a-8d12-41a305b3bf99"},"surfaceOp":"append"} {"type":"step/end","seq":107,"time":1785122256338,"data":{"turn":1,"step":1}} {"type":"step/start","seq":108,"time":1785122256347,"data":{"turn":1,"step":2}} diff --git a/examples/acp-agent/tests/snapshots/cordis-inspect-jsdoc/session.jsonl b/examples/acp-agent/tests/snapshots/cordis-inspect-jsdoc/session.jsonl index 1d41f4aa7e..042753f1a7 100644 --- a/examples/acp-agent/tests/snapshots/cordis-inspect-jsdoc/session.jsonl +++ b/examples/acp-agent/tests/snapshots/cordis-inspect-jsdoc/session.jsonl @@ -1,4 +1,4 @@ -{"type":"session","version":0,"id":"22222222-2222-4222-8222-222222222222","createdAt":1783951000000,"cwd":"/tmp/cordis-inspect-jsdoc","delegationDepth":0} +{"type":"session","version":0,"id":"22222222-2222-4222-8222-222222222222","createdAt":1783951000000,"cwd":"{{cwd}}","delegationDepth":0} {"type":"turn/start","seq":0,"time":1784449176717,"data":{"turn":1,"trigger":{"kind":"message","source":{"kind":"user"}}}} {"type":"user/message","seq":1,"time":1784449176718,"data":{"content":[{"type":"text","text":"Inspect the exact tools service API and tools/pre-execute event with cordis_inspect, then reply with exactly CORDIS_INSPECT_JSDOC_OK."}],"source":{"kind":"user"},"role":"user","id":"48efc8f5-a397-491b-b7a1-179a1185ac2f"},"surfaceOp":"append"} {"type":"session/title","seq":2,"time":1784449176718,"data":{"title":"Inspect the exact tools service","messageSeqs":[1],"source":{"kind":"fallback"}}} diff --git a/examples/acp-agent/tests/snapshots/escalation-approved/session.jsonl b/examples/acp-agent/tests/snapshots/escalation-approved/session.jsonl index 5233644d6e..0985cbd3de 100644 --- a/examples/acp-agent/tests/snapshots/escalation-approved/session.jsonl +++ b/examples/acp-agent/tests/snapshots/escalation-approved/session.jsonl @@ -1,4 +1,4 @@ -{"type":"session","version":0,"id":"f3cbd087-fb45-4b32-b0f2-3082d65bfcb4","createdAt":1783860675270,"cwd":"/var/folders/2g/b32ct0qn1d728l_v6tdkjytr0000gn/T/acp-snap-cwd-cbBLh2","delegationDepth":0} +{"type":"session","version":0,"id":"f3cbd087-fb45-4b32-b0f2-3082d65bfcb4","createdAt":1783860675270,"cwd":"{{cwd}}","delegationDepth":0} {"type":"turn/start","seq":0,"time":1783860675271,"data":{"turn":1,"trigger":{"kind":"message","source":{"kind":"user"}}}} {"type":"user/message","seq":1,"time":1784821261714,"data":{"content":[{"type":"text","text":"The sandbox already denied writing /tmp/dsh-escalated.txt earlier (it is outside this workspace). Retry it now exactly once: one single bash call with the command printf 'escalated\\n' > /tmp/dsh-escalated.txt && cat /tmp/dsh-escalated.txt && rm /tmp/dsh-escalated.txt, with sandbox_permissions set to danger-full-access and the justification 'the user asked to write a file outside the workspace'. Do not run it without sandbox_permissions first. I will approve the permission prompt. After the result, reply with the single word DONE and stop."}],"source":{"kind":"user"},"role":"user","id":"8fcf378f-b720-4a86-be32-95ddec1651c3"},"surfaceOp":"append"} {"type":"session/title","seq":2,"time":1784821261714,"data":{"title":"The sandbox already denied writing","messageSeqs":[1],"source":{"kind":"fallback"}}} diff --git a/examples/acp-agent/tests/snapshots/escalation-rejected/session.jsonl b/examples/acp-agent/tests/snapshots/escalation-rejected/session.jsonl index 85eda59e85..35aebd255b 100644 --- a/examples/acp-agent/tests/snapshots/escalation-rejected/session.jsonl +++ b/examples/acp-agent/tests/snapshots/escalation-rejected/session.jsonl @@ -1,4 +1,4 @@ -{"type":"session","version":0,"id":"d692fe7f-7079-4ee4-8b06-f44fd026d4ea","createdAt":1783860679475,"cwd":"/var/folders/2g/b32ct0qn1d728l_v6tdkjytr0000gn/T/acp-snap-cwd-Hn29Od","delegationDepth":0} +{"type":"session","version":0,"id":"d692fe7f-7079-4ee4-8b06-f44fd026d4ea","createdAt":1783860679475,"cwd":"{{cwd}}","delegationDepth":0} {"type":"turn/start","seq":0,"time":1783860679476,"data":{"turn":1,"trigger":{"kind":"message","source":{"kind":"user"}}}} {"type":"user/message","seq":1,"time":1784821263241,"data":{"content":[{"type":"text","text":"The sandbox already denied writing /tmp/dsh-escalated.txt earlier (it is outside this workspace). Retry it now exactly once: one single bash call with the command printf 'escalated\\n' > /tmp/dsh-escalated.txt && cat /tmp/dsh-escalated.txt && rm /tmp/dsh-escalated.txt, with sandbox_permissions set to danger-full-access and the justification 'the user asked to write a file outside the workspace'. Do not run it without sandbox_permissions first. I will reject the permission prompt; after the rejection, do not retry and do not work around it — explain in one short sentence and stop."}],"source":{"kind":"user"},"role":"user","id":"1f206016-2423-4b51-80bb-df15468298c5"},"surfaceOp":"append"} {"type":"session/title","seq":2,"time":1784821263241,"data":{"title":"The sandbox already denied writing","messageSeqs":[1],"source":{"kind":"fallback"}}} diff --git a/examples/acp-agent/tests/snapshots/fs-edit/session.jsonl b/examples/acp-agent/tests/snapshots/fs-edit/session.jsonl index ae51a7db46..784b6c17c4 100644 --- a/examples/acp-agent/tests/snapshots/fs-edit/session.jsonl +++ b/examples/acp-agent/tests/snapshots/fs-edit/session.jsonl @@ -1,4 +1,4 @@ -{"type":"session","version":0,"id":"736c4bd8-41bd-43fb-9030-b4df3b2a4f83","createdAt":1783352084735,"cwd":"/tmp/acp-snap-cwd-0BxHdV","delegationDepth":0} +{"type":"session","version":0,"id":"736c4bd8-41bd-43fb-9030-b4df3b2a4f83","createdAt":1783352084735,"cwd":"{{cwd}}","delegationDepth":0} {"type":"turn/start","seq":0,"time":1783352084740,"data":{"turn":1,"trigger":{"kind":"message","source":{"kind":"user"}}}} {"type":"user/message","seq":1,"time":1783352084740,"data":{"content":[{"type":"text","text":"First use the read tool to read config.txt in the current directory. Then use the edit tool (NOT bash) to replace the literal text DEBUG with RELEASE in that file. Then reply with exactly the single word DONE."}],"source":{"kind":"user"},"role":"user","id":"6b1ee31e-9c1a-41f3-9647-153d6d98e1a5"},"surfaceOp":"append"} {"type":"session/title","seq":2,"time":1783352084740,"data":{"title":"First use the read tool","messageSeqs":[1],"source":{"kind":"fallback"}}} @@ -14,7 +14,7 @@ {"type":"assistant/chunk","seq":68,"time":1783352086057,"data":{"turn":1,"step":1,"chunk":{"type":"finish","reason":{"kind":"tool-calls"}}}} {"type":"assistant/message","seq":69,"time":1783352086059,"data":{"turn":1,"step":1,"message":{"role":"assistant","content":[{"type":"reasoning","text":"The user wants me to:\n1. Read config.txt in the current directory\n2. Use the edit tool to replace DEBUG with RELEASE\n3. Reply with exactly \"DONE\"\n\nLet me start by reading the file."},{"type":"tool-call","id":"call_00_S6krdIDHoSCGWv7WnXX97617","name":"read","arguments":"{\"file_path\": \"config.txt\"}"}],"source":{"kind":"model","provider":"deepseek","model":"deepseek-v4-flash"},"id":"36ebf262-429c-4398-abbc-a197e2522f1d"},"usage":{"inputTokens":2900,"outputTokens":91,"cacheReadTokens":0,"reasoningTokens":46}},"sourceEventSeqs":[5,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20,21,22,23,24,25,26,27,28,29,30,31,32,33,34,35,36,37,38,39,40,41,42,43,44,45,46,47,48,49,50,51,52,53,54,55,56,57,58,59,60,61,62,63,64,65,66,67,68],"surfaceOp":"append"} {"type":"tool/call","seq":70,"time":1783352086059,"data":{"turn":1,"step":1,"callId":"call_00_S6krdIDHoSCGWv7WnXX97617","name":"read","arguments":"{\"file_path\": \"config.txt\"}"}} -{"type":"tool/result","seq":71,"time":1783352086065,"data":{"turn":1,"step":1,"message":{"source":{"kind":"tool","callId":"call_00_S6krdIDHoSCGWv7WnXX97617"},"content":[{"type":"tool-result","toolCallId":"call_00_S6krdIDHoSCGWv7WnXX97617","content":[{"type":"text","text":"/tmp/acp-snap-cwd-0BxHdV/config.txt\nfile\n\n1: mode=DEBUG\n2: level=info\n\n(End of file - total 2 lines)\n"}],"isError":false}],"role":"user","id":"1c3ce978-55ee-4337-a586-084a77ed44e7"}},"sourceEventSeqs":[70],"surfaceOp":"append"} +{"type":"tool/result","seq":71,"time":1783352086065,"data":{"turn":1,"step":1,"message":{"source":{"kind":"tool","callId":"call_00_S6krdIDHoSCGWv7WnXX97617"},"content":[{"type":"tool-result","toolCallId":"call_00_S6krdIDHoSCGWv7WnXX97617","content":[{"type":"text","text":"{{cwd}}/config.txt\nfile\n\n1: mode=DEBUG\n2: level=info\n\n(End of file - total 2 lines)\n"}],"isError":false}],"role":"user","id":"1c3ce978-55ee-4337-a586-084a77ed44e7"}},"sourceEventSeqs":[70],"surfaceOp":"append"} {"type":"step/end","seq":72,"time":1783352086065,"data":{"turn":1,"step":1}} {"type":"step/start","seq":73,"time":1783352086066,"data":{"turn":1,"step":2}} {"type":"assistant/chunk","seq":74,"time":1783352086901,"data":{"turn":1,"step":2,"chunk":{"type":"block-start","index":0,"blockType":"reasoning"}}} @@ -27,7 +27,7 @@ {"type":"assistant/chunk","seq":128,"time":1783352087469,"data":{"turn":1,"step":2,"chunk":{"type":"finish","reason":{"kind":"tool-calls"}}}} {"type":"assistant/message","seq":129,"time":1783352087469,"data":{"turn":1,"step":2,"message":{"role":"assistant","content":[{"type":"reasoning","text":"Now I need to replace \"DEBUG\" with \"RELEASE\" using the edit tool."},{"type":"tool-call","id":"call_00_vOytneZ0XpsLslEEJAxR6398","name":"edit","arguments":"{\"file_path\": \"config.txt\", \"old_string\": \"DEBUG\", \"new_string\": \"RELEASE\"}"}],"source":{"kind":"model","provider":"deepseek","model":"deepseek-v4-flash"},"id":"c9b571e1-3a63-4a97-af3e-41ac1bdc8e24"},"usage":{"inputTokens":241,"outputTokens":98,"cacheReadTokens":2816,"reasoningTokens":18}},"sourceEventSeqs":[74,75,76,77,78,79,80,81,82,83,84,85,86,87,88,89,90,91,92,93,94,95,96,97,98,99,100,101,102,103,104,105,106,107,108,109,110,111,112,113,114,115,116,117,118,119,120,121,122,123,124,125,126,127,128],"surfaceOp":"append"} {"type":"tool/call","seq":130,"time":1783352087469,"data":{"turn":1,"step":2,"callId":"call_00_vOytneZ0XpsLslEEJAxR6398","name":"edit","arguments":"{\"file_path\": \"config.txt\", \"old_string\": \"DEBUG\", \"new_string\": \"RELEASE\"}"}} -{"type":"tool/result","seq":131,"time":1783352087476,"data":{"turn":1,"step":2,"message":{"source":{"kind":"tool","callId":"call_00_vOytneZ0XpsLslEEJAxR6398"},"content":[{"type":"tool-result","toolCallId":"call_00_vOytneZ0XpsLslEEJAxR6398","content":[{"type":"text","text":"The file /private/tmp/acp-snap-cwd-0BxHdV/config.txt has been updated successfully."}],"isError":false}],"role":"user","id":"79abf084-e65e-468c-84aa-2d3550cb50b8"},"meta":{"diffs":[{"path":"config.txt","oldText":"mode=DEBUG\nlevel=info","newText":"mode=RELEASE\nlevel=info"}]}},"sourceEventSeqs":[130],"surfaceOp":"append"} +{"type":"tool/result","seq":131,"time":1783352087476,"data":{"turn":1,"step":2,"message":{"source":{"kind":"tool","callId":"call_00_vOytneZ0XpsLslEEJAxR6398"},"content":[{"type":"tool-result","toolCallId":"call_00_vOytneZ0XpsLslEEJAxR6398","content":[{"type":"text","text":"The file {{cwd}}/config.txt has been updated successfully."}],"isError":false}],"role":"user","id":"79abf084-e65e-468c-84aa-2d3550cb50b8"},"meta":{"diffs":[{"path":"config.txt","oldText":"mode=DEBUG\nlevel=info","newText":"mode=RELEASE\nlevel=info"}]}},"sourceEventSeqs":[130],"surfaceOp":"append"} {"type":"step/end","seq":132,"time":1783352087477,"data":{"turn":1,"step":2}} {"type":"step/start","seq":133,"time":1783352087477,"data":{"turn":1,"step":3}} {"type":"assistant/chunk","seq":134,"time":1783352088286,"data":{"turn":1,"step":3,"chunk":{"type":"block-start","index":0,"blockType":"reasoning"}}} diff --git a/examples/acp-agent/tests/snapshots/fs-escalation-approved/session.jsonl b/examples/acp-agent/tests/snapshots/fs-escalation-approved/session.jsonl index 52084c3db7..79ef6a1131 100644 --- a/examples/acp-agent/tests/snapshots/fs-escalation-approved/session.jsonl +++ b/examples/acp-agent/tests/snapshots/fs-escalation-approved/session.jsonl @@ -1,4 +1,4 @@ -{"type":"session","version":0,"id":"977a4820-f609-4b48-9039-adcdd921c5fe","createdAt":1784045702340,"cwd":"/var/folders/2g/b32ct0qn1d728l_v6tdkjytr0000gn/T/acp-snap-cwd-vmEGzd","delegationDepth":0} +{"type":"session","version":0,"id":"977a4820-f609-4b48-9039-adcdd921c5fe","createdAt":1784045702340,"cwd":"{{cwd}}","delegationDepth":0} {"type":"turn/start","seq":0,"time":1784045702342,"data":{"turn":1,"trigger":{"kind":"message","source":{"kind":"user"}}}} {"type":"user/message","seq":1,"time":1784821264846,"data":{"content":[{"type":"text","text":"Use the write tool (NOT bash) to create escalated.md in the current directory containing exactly the single line: escalated. An equivalent write was denied earlier, so make this one single write call with sandbox_permissions set to danger-full-access and the justification 'the user asked to escalate this write'. Do not call write without sandbox_permissions first. I will approve the permission prompt. After the result, reply with exactly the single word DONE."}],"source":{"kind":"user"},"role":"user","id":"e4d528b4-0dd8-4aa9-853e-3d00f25b31aa"},"surfaceOp":"append"} {"type":"session/title","seq":2,"time":1784821264846,"data":{"title":"Use the write tool (NOT","messageSeqs":[1],"source":{"kind":"fallback"}}} @@ -16,7 +16,7 @@ {"type":"tool/call","seq":87,"time":1784821264893,"data":{"turn":1,"step":1,"callId":"call_00_Fnymmavpr4klMDy4Fdej3227","name":"write","arguments":"{\"file_path\": \"escalated.md\", \"content\": \"escalated\", \"sandbox_permissions\": \"danger-full-access\", \"justification\": \"the user asked to escalate this write\"}"}} {"type":"approval/asked","seq":88,"time":1784821264898,"data":{"id":"aecaceb0-23b7-4cd5-b7a1-17bc431dc35a","toolName":"write","callId":"call_00_Fnymmavpr4klMDy4Fdej3227","reason":"escalate sandbox to danger-full-access: the user asked to escalate this write"}} {"type":"approval/decided","seq":89,"time":1784821264898,"data":{"id":"aecaceb0-23b7-4cd5-b7a1-17bc431dc35a","outcome":"allowed-once"}} -{"type":"tool/result","seq":90,"time":1784821264906,"data":{"turn":1,"step":1,"message":{"source":{"kind":"tool","callId":"call_00_Fnymmavpr4klMDy4Fdej3227"},"content":[{"type":"tool-result","toolCallId":"call_00_Fnymmavpr4klMDy4Fdej3227","content":[{"type":"text","text":"/private/var/folders/2g/b32ct0qn1d728l_v6tdkjytr0000gn/T/acp-snap-cwd-vmEGzd/escalated.md\nfile\n\nCreated file\n"}],"isError":false}],"role":"user","id":"830d87a2-e325-430d-a463-0911e9512bab"},"meta":{"diffs":[]}},"sourceEventSeqs":[87],"surfaceOp":"append"} +{"type":"tool/result","seq":90,"time":1784821264906,"data":{"turn":1,"step":1,"message":{"source":{"kind":"tool","callId":"call_00_Fnymmavpr4klMDy4Fdej3227"},"content":[{"type":"tool-result","toolCallId":"call_00_Fnymmavpr4klMDy4Fdej3227","content":[{"type":"text","text":"{{cwd}}/escalated.md\nfile\n\nCreated file\n"}],"isError":false}],"role":"user","id":"830d87a2-e325-430d-a463-0911e9512bab"},"meta":{"diffs":[]}},"sourceEventSeqs":[87],"surfaceOp":"append"} {"type":"step/end","seq":91,"time":1784821264911,"data":{"turn":1,"step":1}} {"type":"step/start","seq":92,"time":1784821264912,"data":{"turn":1,"step":2}} {"type":"assistant/chunk","seq":93,"time":1784821264916,"data":{"turn":1,"step":2,"chunk":{"type":"block-start","index":0,"blockType":"reasoning"}}} diff --git a/examples/acp-agent/tests/snapshots/fs-policy-reject/session.jsonl b/examples/acp-agent/tests/snapshots/fs-policy-reject/session.jsonl index 65ebc70d21..d934a2d7be 100644 --- a/examples/acp-agent/tests/snapshots/fs-policy-reject/session.jsonl +++ b/examples/acp-agent/tests/snapshots/fs-policy-reject/session.jsonl @@ -1,4 +1,4 @@ -{"type":"session","version":0,"id":"b3292503-2c3d-4677-804d-1ed6802a4bc5","createdAt":1783611702544,"cwd":"/var/folders/2c/psb0_fmx7hbgz558xjt_f0l00000gn/T/acp-snap-cwd-QzoqnB","delegationDepth":0} +{"type":"session","version":0,"id":"b3292503-2c3d-4677-804d-1ed6802a4bc5","createdAt":1783611702544,"cwd":"{{cwd}}","delegationDepth":0} {"type":"turn/start","seq":0,"time":1783611702550,"data":{"turn":1,"trigger":{"kind":"message","source":{"kind":"user"}}}} {"type":"user/message","seq":1,"time":1783611702550,"data":{"content":[{"type":"text","text":"Do NOT use the read tool and do NOT use bash or shell commands. Immediately use the edit tool to replace the literal text blue with green in settings.txt in the current directory. Do not read the file first. After the tool result, reply with exactly the single word DONE."}],"source":{"kind":"user"},"role":"user","id":"c367f2cd-f9b5-44a4-a363-fdb97d469ad2"},"surfaceOp":"append"} {"type":"session/title","seq":2,"time":1783611702550,"data":{"title":"Do NOT use the read","messageSeqs":[1],"source":{"kind":"fallback"}}} @@ -14,7 +14,7 @@ {"type":"assistant/chunk","seq":76,"time":1783611703969,"data":{"turn":1,"step":1,"chunk":{"type":"finish","reason":{"kind":"tool-calls"}}}} {"type":"assistant/message","seq":77,"time":1783611703972,"data":{"turn":1,"step":1,"message":{"role":"assistant","content":[{"type":"reasoning","text":"The user wants me to use the edit tool to replace \"blue\" with \"green\" in settings.txt without reading the file first, and then reply with just \"DONE\"."},{"type":"tool-call","id":"call_00_x0zlnXl5JOxLrAYL9y7P0119","name":"edit","arguments":"{\"file_path\": \"settings.txt\", \"old_string\": \"blue\", \"new_string\": \"green\"}"}],"source":{"kind":"model","provider":"deepseek","model":"deepseek-v4-flash"},"id":"db6924d3-7ca0-4a50-9bec-9f976b1f493d"},"usage":{"inputTokens":3132,"outputTokens":115,"cacheReadTokens":0,"reasoningTokens":36}},"sourceEventSeqs":[5,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20,21,22,23,24,25,26,27,28,29,30,31,32,33,34,35,36,37,38,39,40,41,42,43,44,45,46,47,48,49,50,51,52,53,54,55,56,57,58,59,60,61,62,63,64,65,66,67,68,69,70,71,72,73,74,75,76],"surfaceOp":"append"} {"type":"tool/call","seq":78,"time":1783611703972,"data":{"turn":1,"step":1,"callId":"call_00_x0zlnXl5JOxLrAYL9y7P0119","name":"edit","arguments":"{\"file_path\": \"settings.txt\", \"old_string\": \"blue\", \"new_string\": \"green\"}"}} -{"type":"tool/result","seq":79,"time":1783611703978,"data":{"turn":1,"step":1,"message":{"source":{"kind":"tool","callId":"call_00_x0zlnXl5JOxLrAYL9y7P0119"},"content":[{"type":"tool-result","toolCallId":"call_00_x0zlnXl5JOxLrAYL9y7P0119","content":[{"type":"text","text":"Error: edit requires reading \"/private/var/folders/2c/psb0_fmx7hbgz558xjt_f0l00000gn/T/acp-snap-cwd-QzoqnB/settings.txt\" first"}],"isError":true}],"role":"user","id":"787330b6-f223-41d6-831e-ce2b14d0e820"},"error":{"name":"FsError","code":"FS_NOT_OBSERVED"}},"sourceEventSeqs":[78],"surfaceOp":"append"} +{"type":"tool/result","seq":79,"time":1783611703978,"data":{"turn":1,"step":1,"message":{"source":{"kind":"tool","callId":"call_00_x0zlnXl5JOxLrAYL9y7P0119"},"content":[{"type":"tool-result","toolCallId":"call_00_x0zlnXl5JOxLrAYL9y7P0119","content":[{"type":"text","text":"Error: edit requires reading \"{{cwd}}/settings.txt\" first"}],"isError":true}],"role":"user","id":"787330b6-f223-41d6-831e-ce2b14d0e820"},"error":{"name":"FsError","code":"FS_NOT_OBSERVED"}},"sourceEventSeqs":[78],"surfaceOp":"append"} {"type":"step/end","seq":80,"time":1783611703978,"data":{"turn":1,"step":1}} {"type":"step/start","seq":81,"time":1783611703978,"data":{"turn":1,"step":2}} {"type":"assistant/chunk","seq":82,"time":1783611704825,"data":{"turn":1,"step":2,"chunk":{"type":"block-start","index":0,"blockType":"reasoning"}}} @@ -27,7 +27,7 @@ {"type":"assistant/chunk","seq":143,"time":1783611705572,"data":{"turn":1,"step":2,"chunk":{"type":"finish","reason":{"kind":"tool-calls"}}}} {"type":"assistant/message","seq":144,"time":1783611705573,"data":{"turn":1,"step":2,"message":{"role":"assistant","content":[{"type":"reasoning","text":"The edit tool requires reading the file first. The user said \"Do not read the file first\" but the tool's own constraints require reading. I need to follow the tool requirements. Let me read the file first."},{"type":"tool-call","id":"call_00_tfr0gdIclW92oIEgi7NF0891","name":"read","arguments":"{\"file_path\": \"settings.txt\"}"}],"source":{"kind":"model","provider":"deepseek","model":"deepseek-v4-flash"},"id":"de588e3c-b10c-4eee-93a5-26e9a665dcbc"},"usage":{"inputTokens":235,"outputTokens":89,"cacheReadTokens":3072,"reasoningTokens":44}},"sourceEventSeqs":[82,83,84,85,86,87,88,89,90,91,92,93,94,95,96,97,98,99,100,101,102,103,104,105,106,107,108,109,110,111,112,113,114,115,116,117,118,119,120,121,122,123,124,125,126,127,128,129,130,131,132,133,134,135,136,137,138,139,140,141,142,143],"surfaceOp":"append"} {"type":"tool/call","seq":145,"time":1783611705573,"data":{"turn":1,"step":2,"callId":"call_00_tfr0gdIclW92oIEgi7NF0891","name":"read","arguments":"{\"file_path\": \"settings.txt\"}"}} -{"type":"tool/result","seq":146,"time":1783611705579,"data":{"turn":1,"step":2,"message":{"source":{"kind":"tool","callId":"call_00_tfr0gdIclW92oIEgi7NF0891"},"content":[{"type":"tool-result","toolCallId":"call_00_tfr0gdIclW92oIEgi7NF0891","content":[{"type":"text","text":"/var/folders/2c/psb0_fmx7hbgz558xjt_f0l00000gn/T/acp-snap-cwd-QzoqnB/settings.txt\nfile\n\n1: color: blue\n\n(End of file - total 1 lines)\n"}],"isError":false}],"role":"user","id":"fade9382-14c7-47e3-8da4-286008d7e9b8"}},"sourceEventSeqs":[145],"surfaceOp":"append"} +{"type":"tool/result","seq":146,"time":1783611705579,"data":{"turn":1,"step":2,"message":{"source":{"kind":"tool","callId":"call_00_tfr0gdIclW92oIEgi7NF0891"},"content":[{"type":"tool-result","toolCallId":"call_00_tfr0gdIclW92oIEgi7NF0891","content":[{"type":"text","text":"{{cwd}}/settings.txt\nfile\n\n1: color: blue\n\n(End of file - total 1 lines)\n"}],"isError":false}],"role":"user","id":"fade9382-14c7-47e3-8da4-286008d7e9b8"}},"sourceEventSeqs":[145],"surfaceOp":"append"} {"type":"step/end","seq":147,"time":1783611705579,"data":{"turn":1,"step":2}} {"type":"step/start","seq":148,"time":1783611705579,"data":{"turn":1,"step":3}} {"type":"assistant/chunk","seq":149,"time":1783611706200,"data":{"turn":1,"step":3,"chunk":{"type":"block-start","index":0,"blockType":"reasoning"}}} @@ -40,7 +40,7 @@ {"type":"assistant/chunk","seq":224,"time":1783611707096,"data":{"turn":1,"step":3,"chunk":{"type":"finish","reason":{"kind":"tool-calls"}}}} {"type":"assistant/message","seq":225,"time":1783611707097,"data":{"turn":1,"step":3,"message":{"role":"assistant","content":[{"type":"reasoning","text":"The file contains \"color: blue\". I need to replace \"blue\" with \"green\". The edit tool said it requires reading first — now I've read it, so the edit should work."},{"type":"tool-call","id":"call_00_GVknJu2tksKkP4lALCwh0926","name":"edit","arguments":"{\"file_path\": \"settings.txt\", \"old_string\": \"blue\", \"new_string\": \"green\"}"}],"source":{"kind":"model","provider":"deepseek","model":"deepseek-v4-flash"},"id":"0e331a4c-4e8d-4daf-a909-80d78a03bdf7"},"usage":{"inputTokens":281,"outputTokens":119,"cacheReadTokens":3200,"reasoningTokens":40}},"sourceEventSeqs":[149,150,151,152,153,154,155,156,157,158,159,160,161,162,163,164,165,166,167,168,169,170,171,172,173,174,175,176,177,178,179,180,181,182,183,184,185,186,187,188,189,190,191,192,193,194,195,196,197,198,199,200,201,202,203,204,205,206,207,208,209,210,211,212,213,214,215,216,217,218,219,220,221,222,223,224],"surfaceOp":"append"} {"type":"tool/call","seq":226,"time":1783611707097,"data":{"turn":1,"step":3,"callId":"call_00_GVknJu2tksKkP4lALCwh0926","name":"edit","arguments":"{\"file_path\": \"settings.txt\", \"old_string\": \"blue\", \"new_string\": \"green\"}"}} -{"type":"tool/result","seq":227,"time":1783611707114,"data":{"turn":1,"step":3,"message":{"source":{"kind":"tool","callId":"call_00_GVknJu2tksKkP4lALCwh0926"},"content":[{"type":"tool-result","toolCallId":"call_00_GVknJu2tksKkP4lALCwh0926","content":[{"type":"text","text":"The file /private/var/folders/2c/psb0_fmx7hbgz558xjt_f0l00000gn/T/acp-snap-cwd-QzoqnB/settings.txt has been updated successfully."}],"isError":false}],"role":"user","id":"e431a509-587b-49fa-8c84-7a6c92e2a014"},"meta":{"diffs":[{"path":"settings.txt","oldText":"color: blue","newText":"color: green"}]}},"sourceEventSeqs":[226],"surfaceOp":"append"} +{"type":"tool/result","seq":227,"time":1783611707114,"data":{"turn":1,"step":3,"message":{"source":{"kind":"tool","callId":"call_00_GVknJu2tksKkP4lALCwh0926"},"content":[{"type":"tool-result","toolCallId":"call_00_GVknJu2tksKkP4lALCwh0926","content":[{"type":"text","text":"The file {{cwd}}/settings.txt has been updated successfully."}],"isError":false}],"role":"user","id":"e431a509-587b-49fa-8c84-7a6c92e2a014"},"meta":{"diffs":[{"path":"settings.txt","oldText":"color: blue","newText":"color: green"}]}},"sourceEventSeqs":[226],"surfaceOp":"append"} {"type":"step/end","seq":228,"time":1783611707114,"data":{"turn":1,"step":3}} {"type":"step/start","seq":229,"time":1783611707114,"data":{"turn":1,"step":4}} {"type":"assistant/chunk","seq":230,"time":1783611707747,"data":{"turn":1,"step":4,"chunk":{"type":"block-start","index":0,"blockType":"reasoning"}}} diff --git a/examples/acp-agent/tests/snapshots/fs-read-window/session.jsonl b/examples/acp-agent/tests/snapshots/fs-read-window/session.jsonl index de2066dc05..72cb3a5200 100644 --- a/examples/acp-agent/tests/snapshots/fs-read-window/session.jsonl +++ b/examples/acp-agent/tests/snapshots/fs-read-window/session.jsonl @@ -1,4 +1,4 @@ -{"type":"session","version":0,"id":"b5639b9d-99a9-49e4-83da-77e6caa702be","createdAt":1783352099834,"cwd":"/tmp/acp-snap-cwd-N9HCkt","delegationDepth":0} +{"type":"session","version":0,"id":"b5639b9d-99a9-49e4-83da-77e6caa702be","createdAt":1783352099834,"cwd":"{{cwd}}","delegationDepth":0} {"type":"turn/start","seq":0,"time":1783352099838,"data":{"turn":1,"trigger":{"kind":"message","source":{"kind":"user"}}}} {"type":"user/message","seq":1,"time":1783352099839,"data":{"content":[{"type":"text","text":"Use the read tool (NOT bash) with offset 5 and limit 4 to read lines 5 through 8 of big.txt in the current directory. Then reply with exactly the single word DONE."}],"source":{"kind":"user"},"role":"user","id":"80cf70ac-0b37-401a-96d2-c54056300cd4"},"surfaceOp":"append"} {"type":"session/title","seq":2,"time":1783352099839,"data":{"title":"Use the read tool (NOT","messageSeqs":[1],"source":{"kind":"fallback"}}} @@ -14,7 +14,7 @@ {"type":"assistant/chunk","seq":90,"time":1783352101346,"data":{"turn":1,"step":1,"chunk":{"type":"finish","reason":{"kind":"tool-calls"}}}} {"type":"assistant/message","seq":91,"time":1783352101348,"data":{"turn":1,"step":1,"message":{"role":"assistant","content":[{"type":"reasoning","text":"The user wants me to use the read tool with offset 5 and limit 4 to read lines 5 through 8 of big.txt in the current directory. Then reply with exactly the single word DONE.\n\nLet me first check the current directory, then read the file."},{"type":"tool-call","id":"call_00_GIZwZS9a7vhWTFCIc7Z35497","name":"read","arguments":"{\"file_path\": \"big.txt\", \"offset\": 5, \"limit\": 4}"}],"source":{"kind":"model","provider":"deepseek","model":"deepseek-v4-flash"},"id":"5620412c-8fae-4d17-aac4-0801f3b02461"},"usage":{"inputTokens":2894,"outputTokens":133,"cacheReadTokens":0,"reasoningTokens":56}},"sourceEventSeqs":[5,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20,21,22,23,24,25,26,27,28,29,30,31,32,33,34,35,36,37,38,39,40,41,42,43,44,45,46,47,48,49,50,51,52,53,54,55,56,57,58,59,60,61,62,63,64,65,66,67,68,69,70,71,72,73,74,75,76,77,78,79,80,81,82,83,84,85,86,87,88,89,90],"surfaceOp":"append"} {"type":"tool/call","seq":92,"time":1783352101348,"data":{"turn":1,"step":1,"callId":"call_00_GIZwZS9a7vhWTFCIc7Z35497","name":"read","arguments":"{\"file_path\": \"big.txt\", \"offset\": 5, \"limit\": 4}"}} -{"type":"tool/result","seq":93,"time":1783352101353,"data":{"turn":1,"step":1,"message":{"source":{"kind":"tool","callId":"call_00_GIZwZS9a7vhWTFCIc7Z35497"},"content":[{"type":"tool-result","toolCallId":"call_00_GIZwZS9a7vhWTFCIc7Z35497","content":[{"type":"text","text":"/tmp/acp-snap-cwd-N9HCkt/big.txt\nfile\n\n5: line five\n6: line six\n7: line seven\n8: line eight\n\n(Showing lines 5-8 of 10. Use offset=9 to continue.)\n"}],"isError":false}],"role":"user","id":"02513672-93cb-4f70-9ee7-ad19542a5f6b"}},"sourceEventSeqs":[92],"surfaceOp":"append"} +{"type":"tool/result","seq":93,"time":1783352101353,"data":{"turn":1,"step":1,"message":{"source":{"kind":"tool","callId":"call_00_GIZwZS9a7vhWTFCIc7Z35497"},"content":[{"type":"tool-result","toolCallId":"call_00_GIZwZS9a7vhWTFCIc7Z35497","content":[{"type":"text","text":"{{cwd}}/big.txt\nfile\n\n5: line five\n6: line six\n7: line seven\n8: line eight\n\n(Showing lines 5-8 of 10. Use offset=9 to continue.)\n"}],"isError":false}],"role":"user","id":"02513672-93cb-4f70-9ee7-ad19542a5f6b"}},"sourceEventSeqs":[92],"surfaceOp":"append"} {"type":"step/end","seq":94,"time":1783352101353,"data":{"turn":1,"step":1}} {"type":"step/start","seq":95,"time":1783352101354,"data":{"turn":1,"step":2}} {"type":"assistant/chunk","seq":96,"time":1783352102021,"data":{"turn":1,"step":2,"chunk":{"type":"block-start","index":0,"blockType":"reasoning"}}} diff --git a/examples/acp-agent/tests/snapshots/fs-read/session.jsonl b/examples/acp-agent/tests/snapshots/fs-read/session.jsonl index 5f19c1b74b..82adec999d 100644 --- a/examples/acp-agent/tests/snapshots/fs-read/session.jsonl +++ b/examples/acp-agent/tests/snapshots/fs-read/session.jsonl @@ -1,4 +1,4 @@ -{"type":"session","version":0,"id":"a57f852d-d476-4716-a380-8a1116e4d905","createdAt":1783352072464,"cwd":"/tmp/acp-snap-cwd-PEETkS","delegationDepth":0} +{"type":"session","version":0,"id":"a57f852d-d476-4716-a380-8a1116e4d905","createdAt":1783352072464,"cwd":"{{cwd}}","delegationDepth":0} {"type":"turn/start","seq":0,"time":1783352072468,"data":{"turn":1,"trigger":{"kind":"message","source":{"kind":"user"}}}} {"type":"user/message","seq":1,"time":1783352072469,"data":{"content":[{"type":"text","text":"Use the read tool (NOT bash) to read the file greeting.txt in the current directory, then reply with exactly the single word DONE."}],"source":{"kind":"user"},"role":"user","id":"7396fa9a-4068-42a6-b153-2b5ade098d32"},"surfaceOp":"append"} {"type":"session/title","seq":2,"time":1783352072469,"data":{"title":"Use the read tool (NOT","messageSeqs":[1],"source":{"kind":"fallback"}}} @@ -14,7 +14,7 @@ {"type":"assistant/chunk","seq":52,"time":1783352073705,"data":{"turn":1,"step":1,"chunk":{"type":"finish","reason":{"kind":"tool-calls"}}}} {"type":"assistant/message","seq":53,"time":1783352073708,"data":{"turn":1,"step":1,"message":{"role":"assistant","content":[{"type":"reasoning","text":"The user wants me to read the file greeting.txt using the read tool (not bash), then reply with exactly the single word \"DONE\"."},{"type":"tool-call","id":"call_00_hHPZCcivsIkXAGS9jTGy8417","name":"read","arguments":"{\"file_path\": \"greeting.txt\"}"}],"source":{"kind":"model","provider":"deepseek","model":"deepseek-v4-flash"},"id":"5452254c-4843-458c-9732-12fe8b7c1468"},"usage":{"inputTokens":2882,"outputTokens":75,"cacheReadTokens":0,"reasoningTokens":29}},"sourceEventSeqs":[5,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20,21,22,23,24,25,26,27,28,29,30,31,32,33,34,35,36,37,38,39,40,41,42,43,44,45,46,47,48,49,50,51,52],"surfaceOp":"append"} {"type":"tool/call","seq":54,"time":1783352073709,"data":{"turn":1,"step":1,"callId":"call_00_hHPZCcivsIkXAGS9jTGy8417","name":"read","arguments":"{\"file_path\": \"greeting.txt\"}"}} -{"type":"tool/result","seq":55,"time":1783352073717,"data":{"turn":1,"step":1,"message":{"source":{"kind":"tool","callId":"call_00_hHPZCcivsIkXAGS9jTGy8417"},"content":[{"type":"tool-result","toolCallId":"call_00_hHPZCcivsIkXAGS9jTGy8417","content":[{"type":"text","text":"/tmp/acp-snap-cwd-PEETkS/greeting.txt\nfile\n\n1: hello\n\n(End of file - total 1 lines)\n"}],"isError":false}],"role":"user","id":"6dd015cf-8c8b-4fd3-a1b2-fa243d67d8e9"}},"sourceEventSeqs":[54],"surfaceOp":"append"} +{"type":"tool/result","seq":55,"time":1783352073717,"data":{"turn":1,"step":1,"message":{"source":{"kind":"tool","callId":"call_00_hHPZCcivsIkXAGS9jTGy8417"},"content":[{"type":"tool-result","toolCallId":"call_00_hHPZCcivsIkXAGS9jTGy8417","content":[{"type":"text","text":"{{cwd}}/greeting.txt\nfile\n\n1: hello\n\n(End of file - total 1 lines)\n"}],"isError":false}],"role":"user","id":"6dd015cf-8c8b-4fd3-a1b2-fa243d67d8e9"}},"sourceEventSeqs":[54],"surfaceOp":"append"} {"type":"step/end","seq":56,"time":1783352073718,"data":{"turn":1,"step":1}} {"type":"step/start","seq":57,"time":1783352073719,"data":{"turn":1,"step":2}} {"type":"assistant/chunk","seq":58,"time":1783352074666,"data":{"turn":1,"step":2,"chunk":{"type":"block-start","index":0,"blockType":"reasoning"}}} diff --git a/examples/acp-agent/tests/snapshots/fs-write-overwrite/session.jsonl b/examples/acp-agent/tests/snapshots/fs-write-overwrite/session.jsonl index 6da33759a6..e46bcfa17c 100644 --- a/examples/acp-agent/tests/snapshots/fs-write-overwrite/session.jsonl +++ b/examples/acp-agent/tests/snapshots/fs-write-overwrite/session.jsonl @@ -1,4 +1,4 @@ -{"type":"session","version":0,"id":"e04cc262-6c89-4586-88d7-3e919240d735","createdAt":1783352092215,"cwd":"/tmp/acp-snap-cwd-hH2sGY","delegationDepth":0} +{"type":"session","version":0,"id":"e04cc262-6c89-4586-88d7-3e919240d735","createdAt":1783352092215,"cwd":"{{cwd}}","delegationDepth":0} {"type":"turn/start","seq":0,"time":1783352092220,"data":{"turn":1,"trigger":{"kind":"message","source":{"kind":"user"}}}} {"type":"user/message","seq":1,"time":1783352092221,"data":{"content":[{"type":"text","text":"First use the read tool to read data.txt in the current directory. Then use the write tool (NOT bash) to replace its entire contents with exactly the single line: replaced. Then reply with exactly the single word DONE."}],"source":{"kind":"user"},"role":"user","id":"5e890158-f455-445a-b265-e0cd1b18af36"},"surfaceOp":"append"} {"type":"session/title","seq":2,"time":1783352092221,"data":{"title":"First use the read tool","messageSeqs":[1],"source":{"kind":"fallback"}}} @@ -14,7 +14,7 @@ {"type":"assistant/chunk","seq":64,"time":1783352093614,"data":{"turn":1,"step":1,"chunk":{"type":"finish","reason":{"kind":"tool-calls"}}}} {"type":"assistant/message","seq":65,"time":1783352093617,"data":{"turn":1,"step":1,"message":{"role":"assistant","content":[{"type":"reasoning","text":"The user wants me to:\n1. Read data.txt using the read tool\n2. Replace its entire contents with exactly \"replaced\" using the write tool\n3. Reply with exactly \"DONE\""},{"type":"tool-call","id":"call_00_n4eRJuGoxNR07svgNtk82243","name":"read","arguments":"{\"file_path\": \"data.txt\"}"}],"source":{"kind":"model","provider":"deepseek","model":"deepseek-v4-flash"},"id":"00272d0c-8ed0-436a-8d10-4a7091447dfe"},"usage":{"inputTokens":2899,"outputTokens":87,"cacheReadTokens":0,"reasoningTokens":42}},"sourceEventSeqs":[5,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20,21,22,23,24,25,26,27,28,29,30,31,32,33,34,35,36,37,38,39,40,41,42,43,44,45,46,47,48,49,50,51,52,53,54,55,56,57,58,59,60,61,62,63,64],"surfaceOp":"append"} {"type":"tool/call","seq":66,"time":1783352093617,"data":{"turn":1,"step":1,"callId":"call_00_n4eRJuGoxNR07svgNtk82243","name":"read","arguments":"{\"file_path\": \"data.txt\"}"}} -{"type":"tool/result","seq":67,"time":1783352093624,"data":{"turn":1,"step":1,"message":{"source":{"kind":"tool","callId":"call_00_n4eRJuGoxNR07svgNtk82243"},"content":[{"type":"tool-result","toolCallId":"call_00_n4eRJuGoxNR07svgNtk82243","content":[{"type":"text","text":"/tmp/acp-snap-cwd-hH2sGY/data.txt\nfile\n\n1: original contents\n\n(End of file - total 1 lines)\n"}],"isError":false}],"role":"user","id":"c14ef7fe-2bb8-4adf-ab15-5a988fbf5f55"}},"sourceEventSeqs":[66],"surfaceOp":"append"} +{"type":"tool/result","seq":67,"time":1783352093624,"data":{"turn":1,"step":1,"message":{"source":{"kind":"tool","callId":"call_00_n4eRJuGoxNR07svgNtk82243"},"content":[{"type":"tool-result","toolCallId":"call_00_n4eRJuGoxNR07svgNtk82243","content":[{"type":"text","text":"{{cwd}}/data.txt\nfile\n\n1: original contents\n\n(End of file - total 1 lines)\n"}],"isError":false}],"role":"user","id":"c14ef7fe-2bb8-4adf-ab15-5a988fbf5f55"}},"sourceEventSeqs":[66],"surfaceOp":"append"} {"type":"step/end","seq":68,"time":1783352093624,"data":{"turn":1,"step":1}} {"type":"step/start","seq":69,"time":1783352093625,"data":{"turn":1,"step":2}} {"type":"assistant/chunk","seq":70,"time":1783352094455,"data":{"turn":1,"step":2,"chunk":{"type":"block-start","index":0,"blockType":"reasoning"}}} @@ -27,7 +27,7 @@ {"type":"assistant/chunk","seq":113,"time":1783352094988,"data":{"turn":1,"step":2,"chunk":{"type":"finish","reason":{"kind":"tool-calls"}}}} {"type":"assistant/message","seq":114,"time":1783352094988,"data":{"turn":1,"step":2,"message":{"role":"assistant","content":[{"type":"reasoning","text":"The file contains \"original contents\". Now I'll replace it with \"replaced\"."},{"type":"tool-call","id":"call_00_N23EvXjDo4c8enyWpIUq4043","name":"write","arguments":"{\"file_path\": \"data.txt\", \"content\": \"replaced\"}"}],"source":{"kind":"model","provider":"deepseek","model":"deepseek-v4-flash"},"id":"17d77822-5c76-4c09-acb6-8ff891129da8"},"usage":{"inputTokens":228,"outputTokens":79,"cacheReadTokens":2816,"reasoningTokens":17}},"sourceEventSeqs":[70,71,72,73,74,75,76,77,78,79,80,81,82,83,84,85,86,87,88,89,90,91,92,93,94,95,96,97,98,99,100,101,102,103,104,105,106,107,108,109,110,111,112,113],"surfaceOp":"append"} {"type":"tool/call","seq":115,"time":1783352094988,"data":{"turn":1,"step":2,"callId":"call_00_N23EvXjDo4c8enyWpIUq4043","name":"write","arguments":"{\"file_path\": \"data.txt\", \"content\": \"replaced\"}"}} -{"type":"tool/result","seq":116,"time":1783352094995,"data":{"turn":1,"step":2,"message":{"source":{"kind":"tool","callId":"call_00_N23EvXjDo4c8enyWpIUq4043"},"content":[{"type":"tool-result","toolCallId":"call_00_N23EvXjDo4c8enyWpIUq4043","content":[{"type":"text","text":"/private/tmp/acp-snap-cwd-hH2sGY/data.txt\nfile\n\nUpdated file\n"}],"isError":false}],"role":"user","id":"2b85c946-b10f-4317-bbf1-e86e5072a4d0"},"meta":{"diffs":[{"path":"data.txt","oldText":"original contents","newText":"replaced"}]}},"sourceEventSeqs":[115],"surfaceOp":"append"} +{"type":"tool/result","seq":116,"time":1783352094995,"data":{"turn":1,"step":2,"message":{"source":{"kind":"tool","callId":"call_00_N23EvXjDo4c8enyWpIUq4043"},"content":[{"type":"tool-result","toolCallId":"call_00_N23EvXjDo4c8enyWpIUq4043","content":[{"type":"text","text":"{{cwd}}/data.txt\nfile\n\nUpdated file\n"}],"isError":false}],"role":"user","id":"2b85c946-b10f-4317-bbf1-e86e5072a4d0"},"meta":{"diffs":[{"path":"data.txt","oldText":"original contents","newText":"replaced"}]}},"sourceEventSeqs":[115],"surfaceOp":"append"} {"type":"step/end","seq":117,"time":1783352094995,"data":{"turn":1,"step":2}} {"type":"step/start","seq":118,"time":1783352094995,"data":{"turn":1,"step":3}} {"type":"assistant/chunk","seq":119,"time":1783352096090,"data":{"turn":1,"step":3,"chunk":{"type":"block-start","index":0,"blockType":"reasoning"}}} diff --git a/examples/acp-agent/tests/snapshots/fs-write/session.jsonl b/examples/acp-agent/tests/snapshots/fs-write/session.jsonl index 1dcc0af155..8d58e22ecc 100644 --- a/examples/acp-agent/tests/snapshots/fs-write/session.jsonl +++ b/examples/acp-agent/tests/snapshots/fs-write/session.jsonl @@ -1,4 +1,4 @@ -{"type":"session","version":0,"id":"fdcab4d0-e5e4-4a06-9195-be8f7049d67e","createdAt":1783352078749,"cwd":"/tmp/acp-snap-cwd-sNvn5N","delegationDepth":0} +{"type":"session","version":0,"id":"fdcab4d0-e5e4-4a06-9195-be8f7049d67e","createdAt":1783352078749,"cwd":"{{cwd}}","delegationDepth":0} {"type":"turn/start","seq":0,"time":1783352078754,"data":{"turn":1,"trigger":{"kind":"message","source":{"kind":"user"}}}} {"type":"user/message","seq":1,"time":1783352078754,"data":{"content":[{"type":"text","text":"Use the write tool (NOT bash) to create a file named notes.txt in the current directory containing exactly the single line: hello world. Then reply with exactly the single word DONE."}],"source":{"kind":"user"},"role":"user","id":"f11a6473-4b11-4205-a73a-edd879e1ec56"},"surfaceOp":"append"} {"type":"session/title","seq":2,"time":1783352078754,"data":{"title":"Use the write tool (NOT","messageSeqs":[1],"source":{"kind":"fallback"}}} @@ -14,7 +14,7 @@ {"type":"assistant/chunk","seq":61,"time":1783352079886,"data":{"turn":1,"step":1,"chunk":{"type":"finish","reason":{"kind":"tool-calls"}}}} {"type":"assistant/message","seq":62,"time":1783352079888,"data":{"turn":1,"step":1,"message":{"role":"assistant","content":[{"type":"reasoning","text":"The user wants me to create a file named notes.txt with the content \"hello world\" using the write tool, then reply with \"DONE\"."},{"type":"tool-call","id":"call_00_APMUCJJm9lrTSlVbg6dB0185","name":"write","arguments":"{\"file_path\": \"notes.txt\", \"content\": \"hello world\"}"}],"source":{"kind":"model","provider":"deepseek","model":"deepseek-v4-flash"},"id":"6a58e768-f922-4850-832e-bfb43d4ab4fc"},"usage":{"inputTokens":2891,"outputTokens":92,"cacheReadTokens":0,"reasoningTokens":30}},"sourceEventSeqs":[5,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20,21,22,23,24,25,26,27,28,29,30,31,32,33,34,35,36,37,38,39,40,41,42,43,44,45,46,47,48,49,50,51,52,53,54,55,56,57,58,59,60,61],"surfaceOp":"append"} {"type":"tool/call","seq":63,"time":1783352079888,"data":{"turn":1,"step":1,"callId":"call_00_APMUCJJm9lrTSlVbg6dB0185","name":"write","arguments":"{\"file_path\": \"notes.txt\", \"content\": \"hello world\"}"}} -{"type":"tool/result","seq":64,"time":1783352079897,"data":{"turn":1,"step":1,"message":{"source":{"kind":"tool","callId":"call_00_APMUCJJm9lrTSlVbg6dB0185"},"content":[{"type":"tool-result","toolCallId":"call_00_APMUCJJm9lrTSlVbg6dB0185","content":[{"type":"text","text":"/private/tmp/acp-snap-cwd-sNvn5N/notes.txt\nfile\n\nCreated file\n"}],"isError":false}],"role":"user","id":"f5031700-edf6-4f15-9dd2-1ebeecaeb762"},"meta":{"diffs":[]}},"sourceEventSeqs":[63],"surfaceOp":"append"} +{"type":"tool/result","seq":64,"time":1783352079897,"data":{"turn":1,"step":1,"message":{"source":{"kind":"tool","callId":"call_00_APMUCJJm9lrTSlVbg6dB0185"},"content":[{"type":"tool-result","toolCallId":"call_00_APMUCJJm9lrTSlVbg6dB0185","content":[{"type":"text","text":"{{cwd}}/notes.txt\nfile\n\nCreated file\n"}],"isError":false}],"role":"user","id":"f5031700-edf6-4f15-9dd2-1ebeecaeb762"},"meta":{"diffs":[]}},"sourceEventSeqs":[63],"surfaceOp":"append"} {"type":"step/end","seq":65,"time":1783352079898,"data":{"turn":1,"step":1}} {"type":"step/start","seq":66,"time":1783352079899,"data":{"turn":1,"step":2}} {"type":"assistant/chunk","seq":67,"time":1783352080825,"data":{"turn":1,"step":2,"chunk":{"type":"block-start","index":0,"blockType":"reasoning"}}} diff --git a/examples/acp-agent/tests/snapshots/hook-cc-posttool-block/session.jsonl b/examples/acp-agent/tests/snapshots/hook-cc-posttool-block/session.jsonl index b7fd854b5a..a43c898146 100644 --- a/examples/acp-agent/tests/snapshots/hook-cc-posttool-block/session.jsonl +++ b/examples/acp-agent/tests/snapshots/hook-cc-posttool-block/session.jsonl @@ -1,4 +1,4 @@ -{"type":"session","version":0,"id":"669e8682-49fc-4dff-9bc7-6280e283cbe4","createdAt":1783962504097,"cwd":"/var/folders/2c/psb0_fmx7hbgz558xjt_f0l00000gn/T/acp-snap-cwd-WxJGUY","delegationDepth":0} +{"type":"session","version":0,"id":"669e8682-49fc-4dff-9bc7-6280e283cbe4","createdAt":1783962504097,"cwd":"{{cwd}}","delegationDepth":0} {"type":"turn/start","seq":0,"time":1783962504115,"data":{"turn":1,"trigger":{"kind":"message","source":{"kind":"user"}}}} {"type":"user/message","seq":1,"time":1783962504117,"data":{"content":[{"type":"text","text":"Call the bash tool to run exactly: echo HELLO. If the first tool result is rejected, retry that command once. Quote the final tool result verbatim and stop."}],"source":{"kind":"user"},"role":"user","id":"775ddb99-fdd1-404f-ba14-4cc37b6ac2c8"},"surfaceOp":"append"} {"type":"session/title","seq":2,"time":1783962504117,"data":{"title":"Call the bash tool to","messageSeqs":[1],"source":{"kind":"fallback"}}} diff --git a/examples/acp-agent/tests/snapshots/hook-cc-posttool-context/session.jsonl b/examples/acp-agent/tests/snapshots/hook-cc-posttool-context/session.jsonl index 5cb9143a71..8bcd0df48f 100644 --- a/examples/acp-agent/tests/snapshots/hook-cc-posttool-context/session.jsonl +++ b/examples/acp-agent/tests/snapshots/hook-cc-posttool-context/session.jsonl @@ -1,4 +1,4 @@ -{"type":"session","version":0,"id":"0a862642-6652-4916-b88d-b058954ab0c6","createdAt":1783352196657,"cwd":"/tmp/acp-snap-cwd-LEetSL","delegationDepth":0} +{"type":"session","version":0,"id":"0a862642-6652-4916-b88d-b058954ab0c6","createdAt":1783352196657,"cwd":"{{cwd}}","delegationDepth":0} {"type":"turn/start","seq":0,"time":1783352196662,"data":{"turn":1,"trigger":{"kind":"message","source":{"kind":"user"}}}} {"type":"user/message","seq":1,"time":1783352196662,"data":{"content":[{"type":"text","text":"Use the bash tool to run exactly: echo HELLO. Report the tool result you got back verbatim, then stop."}],"source":{"kind":"user"},"role":"user","id":"b3957310-0893-4e41-88b2-715c102b5a9a"},"surfaceOp":"append"} {"type":"session/title","seq":2,"time":1783352196662,"data":{"title":"Use the bash tool to","messageSeqs":[1],"source":{"kind":"fallback"}}} diff --git a/examples/acp-agent/tests/snapshots/hook-cc-pretool-ask/session.jsonl b/examples/acp-agent/tests/snapshots/hook-cc-pretool-ask/session.jsonl index 33df927a74..5e3bdb0217 100644 --- a/examples/acp-agent/tests/snapshots/hook-cc-pretool-ask/session.jsonl +++ b/examples/acp-agent/tests/snapshots/hook-cc-pretool-ask/session.jsonl @@ -1,4 +1,4 @@ -{"type":"session","version":0,"id":"f688431c-01a8-4326-a5c5-1b5f0fd08483","createdAt":1783352171511,"cwd":"/tmp/acp-snap-cwd-iKVciS","delegationDepth":0} +{"type":"session","version":0,"id":"f688431c-01a8-4326-a5c5-1b5f0fd08483","createdAt":1783352171511,"cwd":"{{cwd}}","delegationDepth":0} {"type":"turn/start","seq":0,"time":1783352171519,"data":{"turn":1,"trigger":{"kind":"message","source":{"kind":"user"}}}} {"type":"user/message","seq":1,"time":1783352171520,"data":{"content":[{"type":"text","text":"Use the bash tool to run exactly: echo HELLO. Report the tool result you got back verbatim, then stop."}],"source":{"kind":"user"},"role":"user","id":"40085b3d-6b87-4b86-859e-b34786c9a12f"},"surfaceOp":"append"} {"type":"session/title","seq":2,"time":1783352171520,"data":{"title":"Use the bash tool to","messageSeqs":[1],"source":{"kind":"fallback"}}} diff --git a/examples/acp-agent/tests/snapshots/hook-cc-pretool-deny/session.jsonl b/examples/acp-agent/tests/snapshots/hook-cc-pretool-deny/session.jsonl index b5d38ede9e..b956a3f054 100644 --- a/examples/acp-agent/tests/snapshots/hook-cc-pretool-deny/session.jsonl +++ b/examples/acp-agent/tests/snapshots/hook-cc-pretool-deny/session.jsonl @@ -1,4 +1,4 @@ -{"type":"session","version":0,"id":"ff1c1e99-3bd4-4ef8-a954-80d607d628ba","createdAt":1783352165190,"cwd":"/tmp/acp-snap-cwd-wDnkVo","delegationDepth":0} +{"type":"session","version":0,"id":"ff1c1e99-3bd4-4ef8-a954-80d607d628ba","createdAt":1783352165190,"cwd":"{{cwd}}","delegationDepth":0} {"type":"turn/start","seq":0,"time":1783352165195,"data":{"turn":1,"trigger":{"kind":"message","source":{"kind":"user"}}}} {"type":"user/message","seq":1,"time":1783352165196,"data":{"content":[{"type":"text","text":"Use the bash tool to run exactly: echo HELLO. Report the tool result you got back verbatim, then stop."}],"source":{"kind":"user"},"role":"user","id":"57df50c1-78e1-4b8a-857a-c2ae2192dadd"},"surfaceOp":"append"} {"type":"session/title","seq":2,"time":1783352165196,"data":{"title":"Use the bash tool to","messageSeqs":[1],"source":{"kind":"fallback"}}} diff --git a/examples/acp-agent/tests/snapshots/hook-cc-promptsubmit-context/session.jsonl b/examples/acp-agent/tests/snapshots/hook-cc-promptsubmit-context/session.jsonl index 62f310d322..0aeb20331c 100644 --- a/examples/acp-agent/tests/snapshots/hook-cc-promptsubmit-context/session.jsonl +++ b/examples/acp-agent/tests/snapshots/hook-cc-promptsubmit-context/session.jsonl @@ -1,4 +1,4 @@ -{"type":"session","version":0,"id":"d03c3a83-1238-4e2e-ad9a-b86a61840a40","createdAt":1783352160541,"cwd":"/tmp/acp-snap-cwd-QUDqlk","delegationDepth":0} +{"type":"session","version":0,"id":"d03c3a83-1238-4e2e-ad9a-b86a61840a40","createdAt":1783352160541,"cwd":"{{cwd}}","delegationDepth":0} {"type":"turn/start","seq":0,"time":1783352160545,"data":{"turn":1,"trigger":{"kind":"message","source":{"kind":"user"}}}} {"type":"user/message","seq":1,"time":1785122243327,"data":{"content":[{"type":"text","text":"What is my favorite color? Reply with just the color and stop. Do not use any tools."}],"source":{"kind":"user"},"role":"user","id":"7911469c-1e33-4741-9d32-49ecc6a01f0b"},"surfaceOp":"append"} {"type":"user/message","seq":2,"time":1785122243327,"data":{"content":[{"type":"text","text":"The user has previously stated their favorite color is teal."}],"source":{"kind":"plugin","plugin":"hooks-claude"},"role":"user","id":"7b887c49-97bd-46f9-aea4-c462d385a8ee"},"surfaceOp":"append"} diff --git a/examples/acp-agent/tests/snapshots/hook-cc-stop-continue/session.jsonl b/examples/acp-agent/tests/snapshots/hook-cc-stop-continue/session.jsonl index 92c0dc44da..51068cd22e 100644 --- a/examples/acp-agent/tests/snapshots/hook-cc-stop-continue/session.jsonl +++ b/examples/acp-agent/tests/snapshots/hook-cc-stop-continue/session.jsonl @@ -1,4 +1,4 @@ -{"type":"session","version":0,"id":"eda79fbc-8a1b-4226-b74a-f5f297484747","createdAt":1784522140642,"cwd":"/var/folders/4j/54c8wb496zxfrs1ny_21jbb00000gn/T/acp-snap-cwd-r6rWZp","delegationDepth":0} +{"type":"session","version":0,"id":"eda79fbc-8a1b-4226-b74a-f5f297484747","createdAt":1784522140642,"cwd":"{{cwd}}","delegationDepth":0} {"type":"turn/start","seq":0,"time":1784522140646,"data":{"turn":1,"trigger":{"kind":"message","source":{"kind":"user"}}}} {"type":"user/message","seq":1,"time":1784522140647,"data":{"content":[{"type":"text","text":"Reply with the single word FIRST and stop."}],"source":{"kind":"user"},"role":"user","id":"c63da2f2-916d-42cc-8e6f-c9520e1641cd"},"surfaceOp":"append"} {"type":"session/title","seq":2,"time":1784522140647,"data":{"title":"Reply with the single word","messageSeqs":[1],"source":{"kind":"fallback"}}} diff --git a/examples/acp-agent/tests/snapshots/hook-codex-posttool-block/session.jsonl b/examples/acp-agent/tests/snapshots/hook-codex-posttool-block/session.jsonl index 2ede4564c3..67f277d2cd 100644 --- a/examples/acp-agent/tests/snapshots/hook-codex-posttool-block/session.jsonl +++ b/examples/acp-agent/tests/snapshots/hook-codex-posttool-block/session.jsonl @@ -1,4 +1,4 @@ -{"type":"session","version":0,"id":"01aa6a36-e9c2-42ba-934b-30bec80a1658","createdAt":1783986962232,"cwd":"/var/folders/2c/psb0_fmx7hbgz558xjt_f0l00000gn/T/acp-snap-cwd-x67BsP","delegationDepth":0} +{"type":"session","version":0,"id":"01aa6a36-e9c2-42ba-934b-30bec80a1658","createdAt":1783986962232,"cwd":"{{cwd}}","delegationDepth":0} {"type":"turn/start","seq":0,"time":1783986962235,"data":{"turn":1,"trigger":{"kind":"message","source":{"kind":"user"}}}} {"type":"user/message","seq":1,"time":1783986962235,"data":{"content":[{"type":"text","text":"Call the bash tool exactly once to run: echo HELLO. Whatever tool result comes back, quote it verbatim and stop without calling another tool."}],"source":{"kind":"user"},"role":"user","id":"5a3821d5-de5b-4b9c-85b7-d53dca51af5c"},"surfaceOp":"append"} {"type":"session/title","seq":2,"time":1783986962235,"data":{"title":"Call the bash tool exactly","messageSeqs":[1],"source":{"kind":"fallback"}}} diff --git a/examples/acp-agent/tests/snapshots/hook-codex-posttool-context/session.jsonl b/examples/acp-agent/tests/snapshots/hook-codex-posttool-context/session.jsonl index 1cec7b19c9..b1d297049d 100644 --- a/examples/acp-agent/tests/snapshots/hook-codex-posttool-context/session.jsonl +++ b/examples/acp-agent/tests/snapshots/hook-codex-posttool-context/session.jsonl @@ -1,4 +1,4 @@ -{"type":"session","version":0,"id":"39d8aabe-6457-4a0e-83b7-ee33125a3666","createdAt":1783352228436,"cwd":"/tmp/acp-snap-cwd-VGFtPi","delegationDepth":0} +{"type":"session","version":0,"id":"39d8aabe-6457-4a0e-83b7-ee33125a3666","createdAt":1783352228436,"cwd":"{{cwd}}","delegationDepth":0} {"type":"turn/start","seq":0,"time":1783352228441,"data":{"turn":1,"trigger":{"kind":"message","source":{"kind":"user"}}}} {"type":"user/message","seq":1,"time":1783352228442,"data":{"content":[{"type":"text","text":"Use the bash tool to run exactly: echo HELLO. Report the tool result you got back verbatim, then stop."}],"source":{"kind":"user"},"role":"user","id":"7d8954d3-d4e7-4ca6-ba3d-0c5de95a3ace"},"surfaceOp":"append"} {"type":"session/title","seq":2,"time":1783352228442,"data":{"title":"Use the bash tool to","messageSeqs":[1],"source":{"kind":"fallback"}}} diff --git a/examples/acp-agent/tests/snapshots/hook-codex-pretool-block/session.jsonl b/examples/acp-agent/tests/snapshots/hook-codex-pretool-block/session.jsonl index 3ebaa26b9a..97e59bbe8f 100644 --- a/examples/acp-agent/tests/snapshots/hook-codex-pretool-block/session.jsonl +++ b/examples/acp-agent/tests/snapshots/hook-codex-pretool-block/session.jsonl @@ -1,4 +1,4 @@ -{"type":"session","version":0,"id":"57a74aed-99fc-43bc-a875-6dddebf64d69","createdAt":1783352214599,"cwd":"/tmp/acp-snap-cwd-7Hbu0m","delegationDepth":0} +{"type":"session","version":0,"id":"57a74aed-99fc-43bc-a875-6dddebf64d69","createdAt":1783352214599,"cwd":"{{cwd}}","delegationDepth":0} {"type":"turn/start","seq":0,"time":1783352214604,"data":{"turn":1,"trigger":{"kind":"message","source":{"kind":"user"}}}} {"type":"user/message","seq":1,"time":1783352214605,"data":{"content":[{"type":"text","text":"Use the bash tool to run exactly: echo HELLO. Report the tool result you got back verbatim, then stop."}],"source":{"kind":"user"},"role":"user","id":"8aab0b74-e7e0-4c3c-90a3-19a81f2b9c6a"},"surfaceOp":"append"} {"type":"session/title","seq":2,"time":1783352214605,"data":{"title":"Use the bash tool to","messageSeqs":[1],"source":{"kind":"fallback"}}} diff --git a/examples/acp-agent/tests/snapshots/hook-codex-promptsubmit-context/session.jsonl b/examples/acp-agent/tests/snapshots/hook-codex-promptsubmit-context/session.jsonl index 45d09acac1..5ffc12a991 100644 --- a/examples/acp-agent/tests/snapshots/hook-codex-promptsubmit-context/session.jsonl +++ b/examples/acp-agent/tests/snapshots/hook-codex-promptsubmit-context/session.jsonl @@ -1,4 +1,4 @@ -{"type":"session","version":0,"id":"0bebc0f4-a089-4fde-9b6e-db9532cfd4de","createdAt":1783352209682,"cwd":"/tmp/acp-snap-cwd-aopaZV","delegationDepth":0} +{"type":"session","version":0,"id":"0bebc0f4-a089-4fde-9b6e-db9532cfd4de","createdAt":1783352209682,"cwd":"{{cwd}}","delegationDepth":0} {"type":"turn/start","seq":0,"time":1783352209686,"data":{"turn":1,"trigger":{"kind":"message","source":{"kind":"user"}}}} {"type":"user/message","seq":1,"time":1785122250005,"data":{"content":[{"type":"text","text":"What is my favorite color? Reply with just the color and stop. Do not use any tools."}],"source":{"kind":"user"},"role":"user","id":"ea34d65f-e154-4b2a-bea8-3345fdd96658"},"surfaceOp":"append"} {"type":"user/message","seq":2,"time":1785122250006,"data":{"content":[{"type":"text","text":"The user has previously stated their favorite color is teal."}],"source":{"kind":"plugin","plugin":"hooks-codex"},"role":"user","id":"174d8732-a32f-4eb0-8471-d8b3291a34f2"},"surfaceOp":"append"} diff --git a/examples/acp-agent/tests/snapshots/hook-codex-stop-continue/session.jsonl b/examples/acp-agent/tests/snapshots/hook-codex-stop-continue/session.jsonl index 1a9903ce2a..18d6740b2b 100644 --- a/examples/acp-agent/tests/snapshots/hook-codex-stop-continue/session.jsonl +++ b/examples/acp-agent/tests/snapshots/hook-codex-stop-continue/session.jsonl @@ -1,4 +1,4 @@ -{"type":"session","version":0,"id":"eb17be12-ca8c-46c8-b500-0977e8400208","createdAt":1784522152392,"cwd":"/var/folders/4j/54c8wb496zxfrs1ny_21jbb00000gn/T/acp-snap-cwd-ESgqLu","delegationDepth":0} +{"type":"session","version":0,"id":"eb17be12-ca8c-46c8-b500-0977e8400208","createdAt":1784522152392,"cwd":"{{cwd}}","delegationDepth":0} {"type":"turn/start","seq":0,"time":1784522152397,"data":{"turn":1,"trigger":{"kind":"message","source":{"kind":"user"}}}} {"type":"user/message","seq":1,"time":1784522152397,"data":{"content":[{"type":"text","text":"Reply with the single word FIRST and stop."}],"source":{"kind":"user"},"role":"user","id":"c76f1de4-cf89-4f0f-a861-bc699f579f78"},"surfaceOp":"append"} {"type":"session/title","seq":2,"time":1784522152397,"data":{"title":"Reply with the single word","messageSeqs":[1],"source":{"kind":"fallback"}}} diff --git a/examples/acp-agent/tests/snapshots/multi-turn/session.jsonl b/examples/acp-agent/tests/snapshots/multi-turn/session.jsonl index 2425d401ad..2cc17bdcb1 100644 --- a/examples/acp-agent/tests/snapshots/multi-turn/session.jsonl +++ b/examples/acp-agent/tests/snapshots/multi-turn/session.jsonl @@ -1,4 +1,4 @@ -{"type":"session","version":0,"id":"228b7b82-84ed-49b7-a567-981c03b28c77","createdAt":1783352113760,"cwd":"/tmp/acp-snap-cwd-aN2GRR","delegationDepth":0} +{"type":"session","version":0,"id":"228b7b82-84ed-49b7-a567-981c03b28c77","createdAt":1783352113760,"cwd":"{{cwd}}","delegationDepth":0} {"type":"turn/start","seq":0,"time":1783352113765,"data":{"turn":1,"trigger":{"kind":"message","source":{"kind":"user"}}}} {"type":"user/message","seq":1,"time":1783352113765,"data":{"content":[{"type":"text","text":"Reply with exactly the word: ONE. No tools."}],"source":{"kind":"user"},"role":"user","id":"77c88536-5dcd-423c-b2f1-c432d5f057fd"},"surfaceOp":"append"} {"type":"session/title","seq":2,"time":1783352113765,"data":{"title":"Reply with exactly the word:","messageSeqs":[1],"source":{"kind":"fallback"}}} diff --git a/examples/acp-agent/tests/snapshots/packed-chunks/session.jsonl b/examples/acp-agent/tests/snapshots/packed-chunks/session.jsonl index 602b78691d..76f43e17c4 100644 --- a/examples/acp-agent/tests/snapshots/packed-chunks/session.jsonl +++ b/examples/acp-agent/tests/snapshots/packed-chunks/session.jsonl @@ -1,4 +1,4 @@ -{"type":"session","version":0,"id":"ff1c1e99-3bd4-4ef8-a954-80d607d628ba","createdAt":1783352165190,"cwd":"/tmp/acp-snap-cwd-wDnkVo","delegationDepth":0} +{"type":"session","version":0,"id":"ff1c1e99-3bd4-4ef8-a954-80d607d628ba","createdAt":1783352165190,"cwd":"{{cwd}}","delegationDepth":0} {"type":"turn/start","seq":0,"time":1783352165195,"data":{"turn":1,"trigger":{"kind":"message","source":{"kind":"user"}}}} {"type":"user/message","seq":1,"time":1783352165196,"data":{"content":[{"type":"text","text":"Use the bash tool to run exactly: echo HELLO. Report the tool result you got back verbatim, then stop."}],"source":{"kind":"user"},"role":"user","id":"a597583b-7e90-4d4d-9b6a-bb1ab7617417"},"surfaceOp":"append"} {"type":"session/title","seq":2,"time":1783352165196,"data":{"title":"Use the bash tool to","messageSeqs":[1],"source":{"kind":"fallback"}}} diff --git a/examples/acp-agent/tests/snapshots/session-title-after-turn/session.jsonl b/examples/acp-agent/tests/snapshots/session-title-after-turn/session.jsonl index d3bd535ce1..4c2edbcf5d 100644 --- a/examples/acp-agent/tests/snapshots/session-title-after-turn/session.jsonl +++ b/examples/acp-agent/tests/snapshots/session-title-after-turn/session.jsonl @@ -1,4 +1,4 @@ -{"type":"session","version":0,"id":"session-title-after-turn","createdAt":0,"cwd":"/tmp/session-title-after-turn","delegationDepth":0} +{"type":"session","version":0,"id":"session-title-after-turn","createdAt":0,"cwd":"{{cwd}}","delegationDepth":0} {"type":"turn/start","seq":0,"time":1785222848166,"data":{"turn":1,"trigger":{"kind":"message","source":{"kind":"user"}}}} {"type":"user/message","seq":1,"time":1785222848166,"data":{"content":[{"type":"text","text":"Reply with exactly TITLE_DONE. Do not use tools."}],"source":{"kind":"user"},"role":"user","id":"00000000-0000-4000-8000-000000000001"},"surfaceOp":"append"} {"type":"session/title","seq":2,"time":1785222848166,"data":{"title":"Reply with exactly TITLE_DONE. Do","messageSeqs":[1],"source":{"kind":"fallback"}}} diff --git a/examples/acp-agent/tests/snapshots/skill-load/session.jsonl b/examples/acp-agent/tests/snapshots/skill-load/session.jsonl index bab6f80e1c..cf5b8e6026 100644 --- a/examples/acp-agent/tests/snapshots/skill-load/session.jsonl +++ b/examples/acp-agent/tests/snapshots/skill-load/session.jsonl @@ -1,4 +1,4 @@ -{"type":"session","version":0,"id":"9eb4181f-2d05-49d3-98fc-3711fe2f5664","createdAt":1783654655599,"cwd":"/var/folders/_g/59jgff8x2gqd39f5vy1wnbfc0000gn/T/acp-snap-cwd-DhYwNW","delegationDepth":0} +{"type":"session","version":0,"id":"9eb4181f-2d05-49d3-98fc-3711fe2f5664","createdAt":1783654655599,"cwd":"{{cwd}}","delegationDepth":0} {"type":"turn/start","seq":0,"time":1783654655602,"data":{"turn":1,"trigger":{"kind":"message","source":{"kind":"user"}}}} {"type":"user/message","seq":1,"time":1783654655603,"data":{"content":[{"type":"text","text":"Load the snapshot-skill skill with the skill tool, then reply DONE."}],"source":{"kind":"user"},"role":"user","id":"9c670f1c-3508-4b98-9cae-21f363652d6e"},"surfaceOp":"append"} {"type":"session/title","seq":2,"time":1783654655603,"data":{"title":"Load the snapshot-skill skill with","messageSeqs":[1],"source":{"kind":"fallback"}}} @@ -15,7 +15,7 @@ {"type":"assistant/chunk","seq":13,"time":1784903324935,"data":{"turn":1,"step":1,"chunk":{"type":"finish","reason":{"kind":"tool-calls"}}}} {"type":"assistant/message","seq":14,"time":1784903324935,"data":{"turn":1,"step":1,"message":{"role":"assistant","content":[{"type":"reasoning","text":"Load the requested skill."},{"type":"tool-call","id":"call_skill_load","name":"skill","arguments":"{\"name\":\"snapshot-skill\"}"}],"source":{"kind":"model","provider":"deepseek","model":"deepseek-v4-flash"},"id":"cc7d430d-d011-4428-8572-0274c6082277"},"usage":{"inputTokens":100,"outputTokens":20,"cacheReadTokens":0,"reasoningTokens":5}},"sourceEventSeqs":[6,7,8,9,10,11,12,13],"surfaceOp":"append"} {"type":"tool/call","seq":15,"time":1784903324936,"data":{"turn":1,"step":1,"callId":"call_skill_load","name":"skill","arguments":"{\"name\":\"snapshot-skill\"}"}} -{"type":"tool/result","seq":16,"time":1784903324944,"data":{"turn":1,"step":1,"message":{"source":{"kind":"tool","callId":"call_skill_load"},"content":[{"type":"tool-result","toolCallId":"call_skill_load","content":[{"type":"text","text":"\n\nBase directory for this skill: /var/folders/_g/59jgff8x2gqd39f5vy1wnbfc0000gn/T/acp-snap-cwd-DhYwNW/.dsh/skills/snapshot-skill\nResolve relative paths mentioned by this skill against the base directory before using them. Load referenced resources only as needed.\n\n\n\nFollow these snapshot-only instructions.\nResolve referenced resources relative to this skill directory.\n\n"}],"isError":false}],"role":"user","id":"57ec1e09-b3ba-44df-8da0-bb16e7a33bd8"}},"sourceEventSeqs":[15],"surfaceOp":"append"} +{"type":"tool/result","seq":16,"time":1784903324944,"data":{"turn":1,"step":1,"message":{"source":{"kind":"tool","callId":"call_skill_load"},"content":[{"type":"tool-result","toolCallId":"call_skill_load","content":[{"type":"text","text":"\n\nBase directory for this skill: {{cwd}}/.dsh/skills/snapshot-skill\nResolve relative paths mentioned by this skill against the base directory before using them. Load referenced resources only as needed.\n\n\n\nFollow these snapshot-only instructions.\nResolve referenced resources relative to this skill directory.\n\n"}],"isError":false}],"role":"user","id":"57ec1e09-b3ba-44df-8da0-bb16e7a33bd8"}},"sourceEventSeqs":[15],"surfaceOp":"append"} {"type":"step/end","seq":17,"time":1784903324944,"data":{"turn":1,"step":1}} {"type":"step/start","seq":18,"time":1784903324952,"data":{"turn":1,"step":2}} {"type":"assistant/chunk","seq":19,"time":1783654655611,"data":{"turn":1,"step":2,"chunk":{"type":"block-start","index":0,"blockType":"reasoning"}}} diff --git a/examples/acp-agent/tests/snapshots/subagent-depth-two-rejection/session.1.jsonl b/examples/acp-agent/tests/snapshots/subagent-depth-two-rejection/session.1.jsonl index c682fec096..9559b5b378 100644 --- a/examples/acp-agent/tests/snapshots/subagent-depth-two-rejection/session.1.jsonl +++ b/examples/acp-agent/tests/snapshots/subagent-depth-two-rejection/session.1.jsonl @@ -1,4 +1,4 @@ -{"type":"session","version":0,"id":"22222222-2222-4222-8222-222222222222","createdAt":1001,"cwd":"/tmp/subagent-depth-two","parentSession":"11111111-1111-4111-8111-111111111111","delegationDepth":1} +{"type":"session","version":0,"id":"22222222-2222-4222-8222-222222222222","createdAt":1001,"cwd":"{{cwd}}","parentSession":"11111111-1111-4111-8111-111111111111","delegationDepth":1} {"type":"turn/start","seq":0,"time":1784540790312,"data":{"turn":1,"trigger":{"kind":"message","source":{"kind":"user"}}}} {"type":"user/message","seq":1,"time":1784540790312,"data":{"content":[{"type":"text","text":"Call subagent once. Ask that child to attempt one further subagent call, then report the result."}],"source":{"kind":"user"},"role":"user","id":"e1664eb5-480b-4987-a0a3-4fcd85ccb04d"},"surfaceOp":"append"} {"type":"session/title","seq":2,"time":1784540790312,"data":{"title":"Call subagent once. Ask that","messageSeqs":[1],"source":{"kind":"fallback"}}} diff --git a/examples/acp-agent/tests/snapshots/subagent-depth-two-rejection/session.2.jsonl b/examples/acp-agent/tests/snapshots/subagent-depth-two-rejection/session.2.jsonl index 16394c6e65..a9493cbac8 100644 --- a/examples/acp-agent/tests/snapshots/subagent-depth-two-rejection/session.2.jsonl +++ b/examples/acp-agent/tests/snapshots/subagent-depth-two-rejection/session.2.jsonl @@ -1,4 +1,4 @@ -{"type":"session","version":0,"id":"33333333-3333-4333-8333-333333333333","createdAt":1002,"cwd":"/tmp/subagent-depth-two","parentSession":"22222222-2222-4222-8222-222222222222","delegationDepth":2} +{"type":"session","version":0,"id":"33333333-3333-4333-8333-333333333333","createdAt":1002,"cwd":"{{cwd}}","parentSession":"22222222-2222-4222-8222-222222222222","delegationDepth":2} {"type":"turn/start","seq":0,"time":1784540790319,"data":{"turn":1,"trigger":{"kind":"message","source":{"kind":"user"}}}} {"type":"user/message","seq":1,"time":1784540790319,"data":{"content":[{"type":"text","text":"Attempt one subagent call beyond the configured cap, then report the rejection."}],"source":{"kind":"user"},"role":"user","id":"9299d7d1-85e0-4e05-93e4-34d2cf6bafc8"},"surfaceOp":"append"} {"type":"session/title","seq":2,"time":1784540790319,"data":{"title":"Attempt one subagent call beyond","messageSeqs":[1],"source":{"kind":"fallback"}}} diff --git a/examples/acp-agent/tests/snapshots/subagent-depth-two-rejection/session.jsonl b/examples/acp-agent/tests/snapshots/subagent-depth-two-rejection/session.jsonl index 9561b6cc4c..ab2d26180c 100644 --- a/examples/acp-agent/tests/snapshots/subagent-depth-two-rejection/session.jsonl +++ b/examples/acp-agent/tests/snapshots/subagent-depth-two-rejection/session.jsonl @@ -1,4 +1,4 @@ -{"type":"session","version":0,"id":"11111111-1111-4111-8111-111111111111","createdAt":1000,"cwd":"/tmp/subagent-depth-two","delegationDepth":0} +{"type":"session","version":0,"id":"11111111-1111-4111-8111-111111111111","createdAt":1000,"cwd":"{{cwd}}","delegationDepth":0} {"type":"turn/start","seq":0,"time":1784540790290,"data":{"turn":1,"trigger":{"kind":"message","source":{"kind":"user"}}}} {"type":"user/message","seq":1,"time":1784540790291,"data":{"content":[{"type":"text","text":"Delegate through two child generations. The depth-two child must attempt one more subagent call and report the rejection."}],"source":{"kind":"user"},"role":"user","id":"f74eb6a3-3869-4b1c-ba3c-5b6db530ac67"},"surfaceOp":"append"} {"type":"session/title","seq":2,"time":1784540790291,"data":{"title":"Delegate through two child generations.","messageSeqs":[1],"source":{"kind":"fallback"}}} diff --git a/examples/acp-agent/tests/snapshots/subagent-fork/session.1.jsonl b/examples/acp-agent/tests/snapshots/subagent-fork/session.1.jsonl index 98299cec2a..6d5b0d9163 100644 --- a/examples/acp-agent/tests/snapshots/subagent-fork/session.1.jsonl +++ b/examples/acp-agent/tests/snapshots/subagent-fork/session.1.jsonl @@ -1,4 +1,4 @@ -{"type":"session","version":0,"id":"ada8966c-9fa3-441b-8721-37ff1e795e6a","createdAt":1783352137161,"cwd":"/tmp/acp-snap-cwd-0HLtcD","parentSession":"96cf59c9-b347-48b9-b234-a5200913ad05","seedLength":38,"delegationDepth":1} +{"type":"session","version":0,"id":"ada8966c-9fa3-441b-8721-37ff1e795e6a","createdAt":1783352137161,"cwd":"{{cwd}}","parentSession":"96cf59c9-b347-48b9-b234-a5200913ad05","seedLength":38,"delegationDepth":1} {"type":"turn/start","seq":0,"time":1783352134837,"data":{"turn":1,"trigger":{"kind":"message","source":{"kind":"user"}}}} {"type":"user/message","seq":1,"time":1783352134838,"data":{"content":[{"type":"text","text":"Remember this fact for later: the project codeword is MARMALADE. Reply with the single word OK and stop. Do not use any tools."}],"source":{"kind":"user"},"role":"user","id":"917c2f1a-be80-4f54-86e8-c94fe6859bdd"},"surfaceOp":"append"} {"type":"session/title","seq":2,"time":1783352134838,"data":{"title":"Remember this fact for later:","messageSeqs":[1],"source":{"kind":"fallback"}}} diff --git a/examples/acp-agent/tests/snapshots/subagent-fork/session.jsonl b/examples/acp-agent/tests/snapshots/subagent-fork/session.jsonl index debea23a22..a0b09e9478 100644 --- a/examples/acp-agent/tests/snapshots/subagent-fork/session.jsonl +++ b/examples/acp-agent/tests/snapshots/subagent-fork/session.jsonl @@ -1,4 +1,4 @@ -{"type":"session","version":0,"id":"96cf59c9-b347-48b9-b234-a5200913ad05","createdAt":1783352134832,"cwd":"/tmp/acp-snap-cwd-0HLtcD","delegationDepth":0} +{"type":"session","version":0,"id":"96cf59c9-b347-48b9-b234-a5200913ad05","createdAt":1783352134832,"cwd":"{{cwd}}","delegationDepth":0} {"type":"turn/start","seq":0,"time":1783352134837,"data":{"turn":1,"trigger":{"kind":"message","source":{"kind":"user"}}}} {"type":"user/message","seq":1,"time":1783352134838,"data":{"content":[{"type":"text","text":"Remember this fact for later: the project codeword is MARMALADE. Reply with the single word OK and stop. Do not use any tools."}],"source":{"kind":"user"},"role":"user","id":"917c2f1a-be80-4f54-86e8-c94fe6859bdd"},"surfaceOp":"append"} {"type":"session/title","seq":2,"time":1783352134838,"data":{"title":"Remember this fact for later:","messageSeqs":[1],"source":{"kind":"fallback"}}} diff --git a/examples/acp-agent/tests/snapshots/subagent-mixed/session.1.jsonl b/examples/acp-agent/tests/snapshots/subagent-mixed/session.1.jsonl index 5b1c5fcf61..fe94af0f52 100644 --- a/examples/acp-agent/tests/snapshots/subagent-mixed/session.1.jsonl +++ b/examples/acp-agent/tests/snapshots/subagent-mixed/session.1.jsonl @@ -1,4 +1,4 @@ -{"type":"session","version":0,"id":"e4aafa18-b9e3-48d0-8aae-6c9b25dcae80","createdAt":1783352145223,"cwd":"/tmp/acp-snap-cwd-i43JSF","parentSession":"959ffdf5-03e2-465e-9482-009b704632dc","delegationDepth":1} +{"type":"session","version":0,"id":"e4aafa18-b9e3-48d0-8aae-6c9b25dcae80","createdAt":1783352145223,"cwd":"{{cwd}}","parentSession":"959ffdf5-03e2-465e-9482-009b704632dc","delegationDepth":1} {"type":"turn/start","seq":0,"time":1783352145224,"data":{"turn":1,"trigger":{"kind":"message","source":{"kind":"user"}}}} {"type":"user/message","seq":1,"time":1783352145224,"data":{"content":[{"type":"text","text":"Reply with exactly the word ALPHA and nothing else."}],"source":{"kind":"user"},"role":"user","id":"214ad816-8421-48ff-b501-ca51716d761f"},"surfaceOp":"append"} {"type":"session/title","seq":2,"time":1783352145224,"data":{"title":"Reply with exactly the word","messageSeqs":[1],"source":{"kind":"fallback"}}} diff --git a/examples/acp-agent/tests/snapshots/subagent-mixed/session.2.jsonl b/examples/acp-agent/tests/snapshots/subagent-mixed/session.2.jsonl index ce3fd5980d..1b9127889d 100644 --- a/examples/acp-agent/tests/snapshots/subagent-mixed/session.2.jsonl +++ b/examples/acp-agent/tests/snapshots/subagent-mixed/session.2.jsonl @@ -1,4 +1,4 @@ -{"type":"session","version":0,"id":"02b3a8dd-1d5e-4866-825f-5fbf5000a632","createdAt":1783352147504,"cwd":"/tmp/acp-snap-cwd-i43JSF","parentSession":"959ffdf5-03e2-465e-9482-009b704632dc","seedLength":32,"delegationDepth":1} +{"type":"session","version":0,"id":"02b3a8dd-1d5e-4866-825f-5fbf5000a632","createdAt":1783352147504,"cwd":"{{cwd}}","parentSession":"959ffdf5-03e2-465e-9482-009b704632dc","seedLength":32,"delegationDepth":1} {"type":"turn/start","seq":0,"time":1783352142834,"data":{"turn":1,"trigger":{"kind":"message","source":{"kind":"user"}}}} {"type":"user/message","seq":1,"time":1783352142834,"data":{"content":[{"type":"text","text":"Remember this fact for later: the project codeword is SAFFRON. Reply with the single word OK and stop. Do not use any tools."}],"source":{"kind":"user"},"role":"user","id":"867b46b8-e2fa-4257-a2b1-a8fa12abe782"},"surfaceOp":"append"} {"type":"session/title","seq":2,"time":1783352142834,"data":{"title":"Remember this fact for later:","messageSeqs":[1],"source":{"kind":"fallback"}}} diff --git a/examples/acp-agent/tests/snapshots/subagent-mixed/session.jsonl b/examples/acp-agent/tests/snapshots/subagent-mixed/session.jsonl index 65320dbaaf..0ee3d0a595 100644 --- a/examples/acp-agent/tests/snapshots/subagent-mixed/session.jsonl +++ b/examples/acp-agent/tests/snapshots/subagent-mixed/session.jsonl @@ -1,4 +1,4 @@ -{"type":"session","version":0,"id":"959ffdf5-03e2-465e-9482-009b704632dc","createdAt":1783352142830,"cwd":"/tmp/acp-snap-cwd-i43JSF","delegationDepth":0} +{"type":"session","version":0,"id":"959ffdf5-03e2-465e-9482-009b704632dc","createdAt":1783352142830,"cwd":"{{cwd}}","delegationDepth":0} {"type":"turn/start","seq":0,"time":1783352142834,"data":{"turn":1,"trigger":{"kind":"message","source":{"kind":"user"}}}} {"type":"user/message","seq":1,"time":1783352142834,"data":{"content":[{"type":"text","text":"Remember this fact for later: the project codeword is SAFFRON. Reply with the single word OK and stop. Do not use any tools."}],"source":{"kind":"user"},"role":"user","id":"867b46b8-e2fa-4257-a2b1-a8fa12abe782"},"surfaceOp":"append"} {"type":"session/title","seq":2,"time":1783352142834,"data":{"title":"Remember this fact for later:","messageSeqs":[1],"source":{"kind":"fallback"}}} diff --git a/examples/acp-agent/tests/snapshots/subagent-multi/session.1.jsonl b/examples/acp-agent/tests/snapshots/subagent-multi/session.1.jsonl index ef7d8fcbd5..ae13a1f327 100644 --- a/examples/acp-agent/tests/snapshots/subagent-multi/session.1.jsonl +++ b/examples/acp-agent/tests/snapshots/subagent-multi/session.1.jsonl @@ -1,4 +1,4 @@ -{"type":"session","version":0,"id":"553f8e92-aac1-4df3-8657-eacbb58f9581","createdAt":1783352127669,"cwd":"/tmp/acp-snap-cwd-28z5Of","parentSession":"14dda109-5728-45ba-a002-7db9543fe50e","delegationDepth":1} +{"type":"session","version":0,"id":"553f8e92-aac1-4df3-8657-eacbb58f9581","createdAt":1783352127669,"cwd":"{{cwd}}","parentSession":"14dda109-5728-45ba-a002-7db9543fe50e","delegationDepth":1} {"type":"turn/start","seq":0,"time":1783352127670,"data":{"turn":1,"trigger":{"kind":"message","source":{"kind":"user"}}}} {"type":"user/message","seq":1,"time":1783352127670,"data":{"content":[{"type":"text","text":"Reply with exactly the word ALPHA and nothing else."}],"source":{"kind":"user"},"role":"user","id":"4088c6ea-4806-4d0a-a5a7-b430ba9fcb7e"},"surfaceOp":"append"} {"type":"session/title","seq":2,"time":1783352127670,"data":{"title":"Reply with exactly the word","messageSeqs":[1],"source":{"kind":"fallback"}}} diff --git a/examples/acp-agent/tests/snapshots/subagent-multi/session.2.jsonl b/examples/acp-agent/tests/snapshots/subagent-multi/session.2.jsonl index f1c2813380..6939aef6c0 100644 --- a/examples/acp-agent/tests/snapshots/subagent-multi/session.2.jsonl +++ b/examples/acp-agent/tests/snapshots/subagent-multi/session.2.jsonl @@ -1,4 +1,4 @@ -{"type":"session","version":0,"id":"5f49e80c-16fc-42c7-a617-0b6bd0680aa3","createdAt":1783352129662,"cwd":"/tmp/acp-snap-cwd-28z5Of","parentSession":"14dda109-5728-45ba-a002-7db9543fe50e","delegationDepth":1} +{"type":"session","version":0,"id":"5f49e80c-16fc-42c7-a617-0b6bd0680aa3","createdAt":1783352129662,"cwd":"{{cwd}}","parentSession":"14dda109-5728-45ba-a002-7db9543fe50e","delegationDepth":1} {"type":"turn/start","seq":0,"time":1783352129662,"data":{"turn":1,"trigger":{"kind":"message","source":{"kind":"user"}}}} {"type":"user/message","seq":1,"time":1783352129662,"data":{"content":[{"type":"text","text":"Reply with exactly the word BETA and nothing else."}],"source":{"kind":"user"},"role":"user","id":"804b9ed3-e2ed-495e-9840-8e0f657661fe"},"surfaceOp":"append"} {"type":"session/title","seq":2,"time":1783352129662,"data":{"title":"Reply with exactly the word","messageSeqs":[1],"source":{"kind":"fallback"}}} diff --git a/examples/acp-agent/tests/snapshots/subagent-multi/session.jsonl b/examples/acp-agent/tests/snapshots/subagent-multi/session.jsonl index 3179edb205..303ceb6e6b 100644 --- a/examples/acp-agent/tests/snapshots/subagent-multi/session.jsonl +++ b/examples/acp-agent/tests/snapshots/subagent-multi/session.jsonl @@ -1,4 +1,4 @@ -{"type":"session","version":0,"id":"14dda109-5728-45ba-a002-7db9543fe50e","createdAt":1783352126247,"cwd":"/tmp/acp-snap-cwd-28z5Of","delegationDepth":0} +{"type":"session","version":0,"id":"14dda109-5728-45ba-a002-7db9543fe50e","createdAt":1783352126247,"cwd":"{{cwd}}","delegationDepth":0} {"type":"turn/start","seq":0,"time":1783352126251,"data":{"turn":1,"trigger":{"kind":"message","source":{"kind":"user"}}}} {"type":"user/message","seq":1,"time":1783352126251,"data":{"content":[{"type":"text","text":"Use the subagent tool TWICE, once at a time, to delegate two subtasks to child agents. First subtask: 'Reply with exactly the word ALPHA and nothing else.' Second subtask (after the first returns): 'Reply with exactly the word BETA and nothing else.' After both subagents return, reply with the single word PARENT_DONE and stop. Do not use the bash tool."}],"source":{"kind":"user"},"role":"user","id":"26ff1621-20b5-4c1e-b546-ed4c6f6ec99e"},"surfaceOp":"append"} {"type":"session/title","seq":2,"time":1783352126251,"data":{"title":"Use the subagent tool TWICE,","messageSeqs":[1],"source":{"kind":"fallback"}}} diff --git a/examples/acp-agent/tests/snapshots/subagent-spawn/session.1.jsonl b/examples/acp-agent/tests/snapshots/subagent-spawn/session.1.jsonl index 6b107d1c1c..38534a09cf 100644 --- a/examples/acp-agent/tests/snapshots/subagent-spawn/session.1.jsonl +++ b/examples/acp-agent/tests/snapshots/subagent-spawn/session.1.jsonl @@ -1,4 +1,4 @@ -{"type":"session","version":0,"id":"ea339828-7885-42e1-9083-4355e6f1708d","createdAt":1783352120855,"cwd":"/tmp/acp-snap-cwd-rbeWyt","parentSession":"5138ed0d-e86e-4a7d-b75b-803307e92b17","delegationDepth":1} +{"type":"session","version":0,"id":"ea339828-7885-42e1-9083-4355e6f1708d","createdAt":1783352120855,"cwd":"{{cwd}}","parentSession":"5138ed0d-e86e-4a7d-b75b-803307e92b17","delegationDepth":1} {"type":"turn/start","seq":0,"time":1783352120856,"data":{"turn":1,"trigger":{"kind":"message","source":{"kind":"user"}}}} {"type":"user/message","seq":1,"time":1783352120856,"data":{"content":[{"type":"text","text":"Reply with exactly the word CHILD_OK and nothing else."}],"source":{"kind":"user"},"role":"user","id":"f3a2e52a-cfc3-4f9a-b25a-cb48f61e598e"},"surfaceOp":"append"} {"type":"session/title","seq":2,"time":1783352120856,"data":{"title":"Reply with exactly the word","messageSeqs":[1],"source":{"kind":"fallback"}}} diff --git a/examples/acp-agent/tests/snapshots/subagent-spawn/session.jsonl b/examples/acp-agent/tests/snapshots/subagent-spawn/session.jsonl index 51f0899888..f35595a402 100644 --- a/examples/acp-agent/tests/snapshots/subagent-spawn/session.jsonl +++ b/examples/acp-agent/tests/snapshots/subagent-spawn/session.jsonl @@ -1,4 +1,4 @@ -{"type":"session","version":0,"id":"5138ed0d-e86e-4a7d-b75b-803307e92b17","createdAt":1783352119267,"cwd":"/tmp/acp-snap-cwd-rbeWyt","delegationDepth":0} +{"type":"session","version":0,"id":"5138ed0d-e86e-4a7d-b75b-803307e92b17","createdAt":1783352119267,"cwd":"{{cwd}}","delegationDepth":0} {"type":"turn/start","seq":0,"time":1783352119273,"data":{"turn":1,"trigger":{"kind":"message","source":{"kind":"user"}}}} {"type":"user/message","seq":1,"time":1783352119274,"data":{"content":[{"type":"text","text":"Use the subagent tool exactly once to delegate this subtask to a child agent: 'Reply with exactly the word CHILD_OK and nothing else.' After the subagent returns, reply with the single word PARENT_DONE and stop. Do not use the bash tool."}],"source":{"kind":"user"},"role":"user","id":"36c0b82b-ab96-4985-9b44-8895eeedd725"},"surfaceOp":"append"} {"type":"session/title","seq":2,"time":1783352119274,"data":{"title":"Use the subagent tool exactly","messageSeqs":[1],"source":{"kind":"fallback"}}} diff --git a/examples/acp-agent/tests/snapshots/text-turn/session.jsonl b/examples/acp-agent/tests/snapshots/text-turn/session.jsonl index 56aa46d2fa..348c891312 100644 --- a/examples/acp-agent/tests/snapshots/text-turn/session.jsonl +++ b/examples/acp-agent/tests/snapshots/text-turn/session.jsonl @@ -1,4 +1,4 @@ -{"type":"session","version":0,"id":"539aa64c-7f37-40ff-abd8-ed45b717be1b","createdAt":1783600629539,"cwd":"/var/folders/bn/vj1dvck95yd5jh3x4wskflxm0000gn/T/acp-snap-cwd-ka5r8w","delegationDepth":0} +{"type":"session","version":0,"id":"539aa64c-7f37-40ff-abd8-ed45b717be1b","createdAt":1783600629539,"cwd":"{{cwd}}","delegationDepth":0} {"type":"turn/start","seq":0,"time":1783600629541,"data":{"turn":1,"trigger":{"kind":"message","source":{"kind":"user"}}}} {"type":"user/message","seq":1,"time":1783600629541,"data":{"content":[{"type":"text","text":"Reply with exactly the word: PONG. Do not use any tools."}],"source":{"kind":"user"},"role":"user","id":"2da6fcd7-2410-460a-bb8f-bc6491f7b0b0"},"surfaceOp":"append"} {"type":"session/title","seq":2,"time":1783600629541,"data":{"title":"Reply with exactly the word:","messageSeqs":[1],"source":{"kind":"fallback"}}} diff --git a/examples/acp-agent/tests/snapshots/todo-write/session.jsonl b/examples/acp-agent/tests/snapshots/todo-write/session.jsonl index df70a431bd..429c76226e 100644 --- a/examples/acp-agent/tests/snapshots/todo-write/session.jsonl +++ b/examples/acp-agent/tests/snapshots/todo-write/session.jsonl @@ -1,4 +1,4 @@ -{"type":"session","version":0,"id":"b0f1f758-dcf0-474e-851d-e62c11ec0a09","createdAt":1783352057652,"cwd":"/tmp/acp-snap-cwd-AYilT7","delegationDepth":0} +{"type":"session","version":0,"id":"b0f1f758-dcf0-474e-851d-e62c11ec0a09","createdAt":1783352057652,"cwd":"{{cwd}}","delegationDepth":0} {"type":"turn/start","seq":0,"time":1783352057655,"data":{"turn":1,"trigger":{"kind":"message","source":{"kind":"user"}}}} {"type":"user/message","seq":1,"time":1783352057655,"data":{"content":[{"type":"text","text":"Use the todo_write tool to record a plan with exactly three todos: \"read the code\" (in_progress), \"write the fix\" (pending), \"run the tests\" (pending). Send all three in one todo_write call. Then reply with the single word DONE and stop."}],"source":{"kind":"user"},"role":"user","id":"18c389cb-ab26-4a60-96aa-a1314eab3759"},"surfaceOp":"append"} {"type":"session/title","seq":2,"time":1783352057655,"data":{"title":"Use the todo_write tool to","messageSeqs":[1],"source":{"kind":"fallback"}}} diff --git a/examples/acp-agent/tests/snapshots/tool-call-turn/session.jsonl b/examples/acp-agent/tests/snapshots/tool-call-turn/session.jsonl index 9107893db5..24ea03494f 100644 --- a/examples/acp-agent/tests/snapshots/tool-call-turn/session.jsonl +++ b/examples/acp-agent/tests/snapshots/tool-call-turn/session.jsonl @@ -1,4 +1,4 @@ -{"type":"session","version":0,"id":"e9421ff4-baae-4807-a7ea-fd8a65f2c897","createdAt":1783352044766,"cwd":"/tmp/acp-snap-cwd-OwUkBh","delegationDepth":0} +{"type":"session","version":0,"id":"e9421ff4-baae-4807-a7ea-fd8a65f2c897","createdAt":1783352044766,"cwd":"{{cwd}}","delegationDepth":0} {"type":"turn/start","seq":0,"time":1783352044771,"data":{"turn":1,"trigger":{"kind":"message","source":{"kind":"user"}}}} {"type":"user/message","seq":1,"time":1783352044771,"data":{"content":[{"type":"text","text":"Use the bash tool to run exactly: echo SNAPSHOT_OK. Then reply with the single word DONE and stop."}],"source":{"kind":"user"},"role":"user","id":"033e6f20-6021-4ecc-a80f-de758a3dc877"},"surfaceOp":"append"} {"type":"session/title","seq":2,"time":1783352044771,"data":{"title":"Use the bash tool to","messageSeqs":[1],"source":{"kind":"fallback"}}} diff --git a/examples/acp-agent/tests/snapshots/web-fetch/session.jsonl b/examples/acp-agent/tests/snapshots/web-fetch/session.jsonl index f92f59bc11..396860773e 100644 --- a/examples/acp-agent/tests/snapshots/web-fetch/session.jsonl +++ b/examples/acp-agent/tests/snapshots/web-fetch/session.jsonl @@ -1,4 +1,4 @@ -{"type":"session","version":0,"id":"c12fa9af-1042-4a92-9ba4-4a968ff23495","createdAt":1785078727712,"cwd":"/tmp/acp-snap-cwd-hqkZWE","delegationDepth":0} +{"type":"session","version":0,"id":"c12fa9af-1042-4a92-9ba4-4a968ff23495","createdAt":1785078727712,"cwd":"{{cwd}}","delegationDepth":0} {"type":"turn/start","seq":0,"time":1785078727718,"data":{"turn":1,"trigger":{"kind":"message","source":{"kind":"user"}}}} {"type":"user/message","seq":1,"time":1785078727719,"data":{"content":[{"type":"text","text":"Use the web_fetch tool exactly once to fetch http://127.0.0.1:43117/menu.html, then reply with exactly DONE. Do not describe the content."}],"source":{"kind":"user"},"role":"user","id":"6c8e9279-bb26-4369-b425-951cd33d6b15"},"surfaceOp":"append"} {"type":"session/title","seq":2,"time":1785078727721,"data":{"title":"Use the web_fetch tool exactly","messageSeqs":[1],"source":{"kind":"fallback"}}} diff --git a/examples/acp-agent/tests/snapshots/workflow-run/session.1.jsonl b/examples/acp-agent/tests/snapshots/workflow-run/session.1.jsonl index 4c00ccc7c1..268c7db0f6 100644 --- a/examples/acp-agent/tests/snapshots/workflow-run/session.1.jsonl +++ b/examples/acp-agent/tests/snapshots/workflow-run/session.1.jsonl @@ -1,4 +1,4 @@ -{"type":"session","version":0,"id":"583a4db2-3350-436c-b4a5-5615fd159052","createdAt":1783600636316,"cwd":"/var/folders/bn/vj1dvck95yd5jh3x4wskflxm0000gn/T/acp-snap-cwd-vdJYjz","parentSession":"3fd7d599-56b1-493a-930d-f1fc5e1556e8","delegationDepth":1} +{"type":"session","version":0,"id":"583a4db2-3350-436c-b4a5-5615fd159052","createdAt":1783600636316,"cwd":"{{cwd}}","parentSession":"3fd7d599-56b1-493a-930d-f1fc5e1556e8","delegationDepth":1} {"type":"turn/start","seq":0,"time":1783600636316,"data":{"turn":1,"trigger":{"kind":"message","source":{"kind":"user"}}}} {"type":"user/message","seq":1,"time":1783600636316,"data":{"content":[{"type":"text","text":"Reply with exactly the word WF_CHILD_OK and nothing else."}],"source":{"kind":"user"},"role":"user","id":"660a2954-67fc-4406-8703-189f3c0ee81e"},"surfaceOp":"append"} {"type":"session/title","seq":2,"time":1783600636316,"data":{"title":"Reply with exactly the word","messageSeqs":[1],"source":{"kind":"fallback"}}} diff --git a/examples/acp-agent/tests/snapshots/workflow-run/session.jsonl b/examples/acp-agent/tests/snapshots/workflow-run/session.jsonl index 700410718c..aa29969cf5 100644 --- a/examples/acp-agent/tests/snapshots/workflow-run/session.jsonl +++ b/examples/acp-agent/tests/snapshots/workflow-run/session.jsonl @@ -1,4 +1,4 @@ -{"type":"session","version":0,"id":"3fd7d599-56b1-493a-930d-f1fc5e1556e8","createdAt":1783600631835,"cwd":"/var/folders/bn/vj1dvck95yd5jh3x4wskflxm0000gn/T/acp-snap-cwd-vdJYjz","delegationDepth":0} +{"type":"session","version":0,"id":"3fd7d599-56b1-493a-930d-f1fc5e1556e8","createdAt":1783600631835,"cwd":"{{cwd}}","delegationDepth":0} {"type":"turn/start","seq":0,"time":1783600631838,"data":{"turn":1,"trigger":{"kind":"message","source":{"kind":"user"}}}} {"type":"user/message","seq":1,"time":1783600631838,"data":{"content":[{"type":"text","text":"Use the workflow tool exactly once, with args omitted, meta set to { \"name\": \"snapshot-flow\", \"description\": \"one child for the snapshot\" }, and this EXACT script body (copy it verbatim):\nphase('Run')\nconst reply = await agent('Reply with exactly the word WF_CHILD_OK and nothing else.')\nreturn { reply }\nAfter the workflow returns, reply with the single word WORKFLOW_DONE and stop. Do not use any other tool."}],"source":{"kind":"user"},"role":"user","id":"7752d242-0fc3-421c-ad28-60333479140c"},"surfaceOp":"append"} {"type":"session/title","seq":2,"time":1783600631838,"data":{"title":"Use the workflow tool exactly","messageSeqs":[1],"source":{"kind":"fallback"}}} diff --git a/examples/acp-agent/tests/snapshots/workspace-edit/session.jsonl b/examples/acp-agent/tests/snapshots/workspace-edit/session.jsonl index e5a133dea3..eaee6bd14b 100644 --- a/examples/acp-agent/tests/snapshots/workspace-edit/session.jsonl +++ b/examples/acp-agent/tests/snapshots/workspace-edit/session.jsonl @@ -1,4 +1,4 @@ -{"type":"session","version":0,"id":"48aca674-000a-4583-810b-01f8785cef13","createdAt":1783352264076,"cwd":"/tmp/acp-snap-cwd-rxbEpP","delegationDepth":0} +{"type":"session","version":0,"id":"48aca674-000a-4583-810b-01f8785cef13","createdAt":1783352264076,"cwd":"{{cwd}}","delegationDepth":0} {"type":"turn/start","seq":0,"time":1783352264080,"data":{"turn":1,"trigger":{"kind":"message","source":{"kind":"user"}}}} {"type":"user/message","seq":1,"time":1783352264081,"data":{"content":[{"type":"text","text":"A file named greeting.txt in the current directory contains one word. Use the bash tool to append a second line containing the word WORLD to it (so it has two lines), then read the file back with `cat greeting.txt` to confirm, and reply with the single word DONE. Use a single bash call per action."}],"source":{"kind":"user"},"role":"user","id":"77ac6781-b796-4060-b670-63baa39a986b"},"surfaceOp":"append"} {"type":"session/title","seq":2,"time":1783352264081,"data":{"title":"A file named greeting.txt in","messageSeqs":[1],"source":{"kind":"fallback"}}} @@ -14,7 +14,7 @@ {"type":"assistant/chunk","seq":78,"time":1783352265489,"data":{"turn":1,"step":1,"chunk":{"type":"finish","reason":{"kind":"tool-calls"}}}} {"type":"assistant/message","seq":79,"time":1783352265491,"data":{"turn":1,"step":1,"message":{"role":"assistant","content":[{"type":"reasoning","text":"The user wants me to:\n1. Read the file greeting.txt\n2. Append the word WORLD as a second line\n3. Read the file back with cat to confirm\n4. Reply with DONE\n\nLet me start by reading the file to see its contents."},{"type":"tool-call","id":"call_00_OjRFB4zvxu6UALDjytZD0978","name":"read","arguments":"{\"file_path\": \"greeting.txt\"}"}],"source":{"kind":"model","provider":"deepseek","model":"deepseek-v4-flash"},"id":"3f154ea9-6cf0-4d0a-a478-503962bfe8e1"},"usage":{"inputTokens":2918,"outputTokens":101,"cacheReadTokens":0,"reasoningTokens":55}},"sourceEventSeqs":[5,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20,21,22,23,24,25,26,27,28,29,30,31,32,33,34,35,36,37,38,39,40,41,42,43,44,45,46,47,48,49,50,51,52,53,54,55,56,57,58,59,60,61,62,63,64,65,66,67,68,69,70,71,72,73,74,75,76,77,78],"surfaceOp":"append"} {"type":"tool/call","seq":80,"time":1783352265491,"data":{"turn":1,"step":1,"callId":"call_00_OjRFB4zvxu6UALDjytZD0978","name":"read","arguments":"{\"file_path\": \"greeting.txt\"}"}} -{"type":"tool/result","seq":81,"time":1783352265504,"data":{"turn":1,"step":1,"message":{"source":{"kind":"tool","callId":"call_00_OjRFB4zvxu6UALDjytZD0978"},"content":[{"type":"tool-result","toolCallId":"call_00_OjRFB4zvxu6UALDjytZD0978","content":[{"type":"text","text":"/tmp/acp-snap-cwd-rxbEpP/greeting.txt\nfile\n\n1: hello\n\n(End of file - total 1 lines)\n"}],"isError":false}],"role":"user","id":"59ffbde4-d450-4564-a907-beeec29af0d0"}},"sourceEventSeqs":[80],"surfaceOp":"append"} +{"type":"tool/result","seq":81,"time":1783352265504,"data":{"turn":1,"step":1,"message":{"source":{"kind":"tool","callId":"call_00_OjRFB4zvxu6UALDjytZD0978"},"content":[{"type":"tool-result","toolCallId":"call_00_OjRFB4zvxu6UALDjytZD0978","content":[{"type":"text","text":"{{cwd}}/greeting.txt\nfile\n\n1: hello\n\n(End of file - total 1 lines)\n"}],"isError":false}],"role":"user","id":"59ffbde4-d450-4564-a907-beeec29af0d0"}},"sourceEventSeqs":[80],"surfaceOp":"append"} {"type":"step/end","seq":82,"time":1783352265504,"data":{"turn":1,"step":1}} {"type":"step/start","seq":83,"time":1783352265505,"data":{"turn":1,"step":2}} {"type":"assistant/chunk","seq":84,"time":1783352266385,"data":{"turn":1,"step":2,"chunk":{"type":"block-start","index":0,"blockType":"reasoning"}}} diff --git a/examples/headless-agent/subagent-inheritance.cordis.snapshot.yml b/examples/headless-agent/subagent-inheritance.cordis.snapshot.yml new file mode 100644 index 0000000000..fb614a5372 --- /dev/null +++ b/examples/headless-agent/subagent-inheritance.cordis.snapshot.yml @@ -0,0 +1,69 @@ +# Keyless real-Loader composition for the parent-only override inheritance +# snapshot. The deployment default stays WIDE (workspace-write) while the +# seeded parent session carries a session-scoped read-only override; the +# resumed parent delegates, and only the inheritance capture can confine the +# child — remove it and the child writes successfully under the deployment +# default, so this scenario is the assembled-app red/green anchor for the +# delegation bypass. + +- id: persistence + name: '@deepseek-ai/dsh-session-persistence-jsonl' + config: + root: './.sessions' + compression: none + +# file/override/childFiles all default to their DSH_SNAPSHOT_* env vars. +- id: replay + name: '@deepseek-ai/dsh-llm-replay' + +- id: bash + name: '@deepseek-ai/dsh-bash-local' + config: + timeoutMs: 60000 + +# The confining filesystem stack: the wide deployment default lives on the +# shared policy home; the seeded parent's read-only override must beat it +# INSIDE the child for the scenario to deny. +- id: sandbox-policy + name: '@deepseek-ai/dsh-sandbox-policy' + config: + mode: workspace-write + workspaceRoot: !!js process.cwd() + +- id: fs-sandbox + name: '@deepseek-ai/dsh-fs-sandbox' + +- id: fs-policy + name: '@deepseek-ai/dsh-fs-policy' + +- id: tool-fs + name: '@deepseek-ai/dsh-tool-fs' + +- id: agent + name: '@deepseek-ai/dsh-agent-spine-demo' + config: + agents: [] + workspaceContext: false + skills: + enabled: false + toolTasks: false + goals: false + +- id: subagent + name: '@deepseek-ai/dsh-subagent' + +- id: subagent-spawn + name: '@deepseek-ai/dsh-subagent-spawn' + config: + providerName: spawn + +- id: tool-subagent + name: '@deepseek-ai/dsh-tool-subagent' + config: + provider: spawn + toolName: subagent + maxDepth: 1 + +# Await the persisted resume before the headless driver inspects root agents. +- id: resumed-agent + name: './tests/fixtures/subagent-inheritance-agent.ts' diff --git a/examples/headless-agent/tests/fixtures/subagent-inheritance-agent.ts b/examples/headless-agent/tests/fixtures/subagent-inheritance-agent.ts new file mode 100644 index 0000000000..bd8a7aa2f7 --- /dev/null +++ b/examples/headless-agent/tests/fixtures/subagent-inheritance-agent.ts @@ -0,0 +1,25 @@ +/** + * Loader fixture that resumes the seeded read-only parent before CLI dispatch. + * @module subagent-inheritance-agent + */ + +import type { Context } from 'cordis' +import type { SessionId } from '@deepseek-ai/dsh-session' + +/** Fixture plugin name. */ +export const name = 'subagent-inheritance-agent' +/** Services that must exist before the fixture resumes its agent. */ +export const inject = ['agents', 'agentLoop', 'sessionPersistence'] + +/** + * Resume the seeded session and bind its exact handle to this fixture's lifetime. + * @param ctx - settled agent and persistence services from the Loader tree. + * @returns after the resumed agent is published. + */ +export async function apply(ctx: Context): Promise { + const handle = await ctx.agents.resume({ + resumeSessionId: 'subagent-inheritance-parent' as SessionId, + agentOptions: { provider: 'deepseek', model: 'deepseek-v4-flash' }, + }) + ctx.effect(() => () => handle.dispose(), 'subagent-inheritance-agent.handle') +} diff --git a/examples/headless-agent/tests/headless.snapshot.ts b/examples/headless-agent/tests/headless.snapshot.ts index 53a8342d93..2581d8a042 100644 --- a/examples/headless-agent/tests/headless.snapshot.ts +++ b/examples/headless-agent/tests/headless.snapshot.ts @@ -7,6 +7,7 @@ import { refreshFixtureReplacements, scrubRequestHeaders, stabilizeRefreshLog, + tokenizeSessionFixtureCwd, type HarvestedLog, type NormalizeContext, } from '@deepseek-ai/dsh-acp-snapshot' @@ -260,7 +261,9 @@ describe('headless stream-json snapshots', () => { if (existing === undefined || file === undefined) { throw new Error(`headless snapshot has no fixture for persisted log ${index}`) } - const stable = stabilizeRefreshLog(actual.content, existing, replacements, actualContext) + const stable = tokenizeSessionFixtureCwd( + stabilizeRefreshLog(actual.content, existing, replacements, actualContext), + ) await writeFile(file, stable) return stable })) @@ -456,7 +459,9 @@ describe('headless stream-json snapshots', () => { content: actual.content, } const replacements = refreshFixtureReplacements([harvested], [expectedSession]) - expectedSession = stabilizeRefreshLog(actual.content, expectedSession, replacements, actualContext) + expectedSession = tokenizeSessionFixtureCwd( + stabilizeRefreshLog(actual.content, expectedSession, replacements, actualContext), + ) await writeFile(ptySessionFixture, expectedSession) } const expectedContext = contextFromLogs([expectedSession]) diff --git a/examples/headless-agent/tests/snapshots/advanced-toolchain/session.1.jsonl b/examples/headless-agent/tests/snapshots/advanced-toolchain/session.1.jsonl index c41a628b57..f412b843ee 100644 --- a/examples/headless-agent/tests/snapshots/advanced-toolchain/session.1.jsonl +++ b/examples/headless-agent/tests/snapshots/advanced-toolchain/session.1.jsonl @@ -1,9 +1,9 @@ -{"type":"session","version":0,"id":"22222222-2222-4222-8222-222222222222","createdAt":1783950001000,"cwd":"/tmp/advanced-headless","parentSession":"11111111-1111-4111-8111-111111111111","delegationDepth":1} +{"type":"session","version":0,"id":"22222222-2222-4222-8222-222222222222","createdAt":1783950001000,"cwd":"{{cwd}}","parentSession":"11111111-1111-4111-8111-111111111111","delegationDepth":1} {"type":"turn/start","seq":0,"time":1783957884563,"data":{"turn":1,"trigger":{"kind":"message","source":{"kind":"user"}}}} {"type":"user/message","seq":1,"time":1783957884563,"data":{"content":[{"type":"text","text":"Reply with exactly DIRECT_CHILD_OK and nothing else."}],"source":{"kind":"user"},"role":"user","id":"b07a1eeb-2060-44e5-87d3-05d315a4a74b"},"surfaceOp":"append"} {"type":"session/title","seq":2,"time":1783957884563,"data":{"title":"Reply with exactly DIRECT_CHILD_OK and","messageSeqs":[1],"source":{"kind":"fallback"}}} {"type":"step/start","seq":3,"time":1783957884564,"data":{"turn":1,"step":1}} -{"type":"request/header","seq":4,"time":1783957884564,"data":{"header":{"config":{"provider":"deepseek","model":"deepseek-v4-flash"},"system":"You are an AI agent powered by the DeepSeek Harness SDK.\n\nYou are headless-agent, a coding assistant powered by the deepseek-v4-flash model. Your working directory is /tmp/advanced-headless.\n\nVerify your work by running the code or tests. Keep answers brief and factual.\n\n\nUse the read tool — not shell commands like cat — to inspect text files. Results include line numbers. Use offset and limit to continue reading large files.\n\nUse the write tool to create files or completely replace file contents. Existing files are overwritten, so read an existing file first (the default fs-policy requires it) and prefer edit for targeted changes.\n\nUse the edit tool for targeted changes to existing UTF-8 text files. It replaces literal old_string with new_string; by default old_string must appear exactly once. If old_string appears multiple times, provide a more specific old_string or set replace_all to true. Read the file first (the default fs-policy requires it), unless you just created or edited it in this session.\n\nCheck the [exit code: N] marker on every bash result; investigate failures before moving on.\n\nTrack every background task id you start. You are notified in-session when a task finishes — do not busy-poll or sleep on one; keep working on independent steps and do not duplicate a running task's work. Before giving a final answer, collect every still-relevant task with task_output (set wait: true only when you are genuinely blocked on it), and task_kill tasks that stopped mattering.\n\nUse the workflow tool ONLY when the user explicitly asks for a workflow or for large multi-agent orchestration: you write a JavaScript script (the tool description documents the exact format) that fans work out across many subagents with phases and structured results. For one or two delegations, prefer plain subagent calls.\n\nUse the ralph tool ONLY when the direct human explicitly asks for a Ralph loop or fresh-agent iterative execution. Each Ralph round starts a fresh child with no conversation seed and uses the shared workspace as durable memory. Completion and blockers are worker reports, not independent evaluation. Use same-session goal tools for ordinary long-running objectives, and plain subagents or workflows for bounded delegation and fan-out.\n\n## Writing code for run_code\n\nPass `run_code` the body of an async TypeScript function (erasable syntax only — no `enum` or namespaces; type annotations are advisory, the code runs type-stripped). Inside the program:\n\n- Call tools as `await tools.name(args)` — quoted access for exotic names: `tools[\"my-tool\"](args)`. Every call resolves to the tool's typed canonical JSON value. Tool arguments must be lossless JSON.\n- A FAILED tool call rejects with `ToolCallError`, whose `toolName` identifies the failed tool and whose `message` is human-readable — `try/catch` it to handle and continue.\n- Independent read-only calls MAY overlap under `Promise.all` (safe calls run concurrently; mutating calls run alone, in submission order). Sequence dependent work with `await`.\n- Emit results with `return` and/or `console.log(...)`. ONLY what you print or return comes back to you — intermediate tool results never enter the conversation, so extract just what you need.\n\nThe available tools:\n\n```ts\ntype JsonValue = null | boolean | number | string | JsonValue[] | { [key: string]: JsonValue }\n\ninterface ToolArgsMap {\n /** Execute a bash command (`bash -c`) and return its stdout/stderr. Each call runs in a fresh shell: no state (cwd, variables, functions) persists between calls — pass `workdir` instead of using `cd`. Non-zero exits are reported as `[exit code: N]`. Current harness environment facts are exposed through managed `$DSH_*` variables; inspect them when needed. Commands may run under a file sandbox; a blocked file operation is reported as `[sandbox: file access denied under mode]` — a policy denial, not a bug in the command; do not retry another way. Long output is truncated to its tail; the full output is saved to a file whose path is reported when available. Set `run_in_background: true` for long-running commands: the call returns a task id immediately; read its output with `task_output` and stop it with `task_kill`. */\n bash: {\n /** The bash command to execute. */\n command: string;\n /** Clear, concise description of what this command does in active voice, 5-10 words (shown in the UI). Examples: \"ls\" → \"List files in current directory\"; \"git status\" → \"Show working tree status\"; \"npm install\" → \"Install package dependencies\". */\n description: string;\n /** Timeout in milliseconds. The executor applies its configured default and cap, and kills the command on expiry. */\n timeoutMs?: number;\n /** Working directory for this command. Defaults to the session workspace; a relative path is resolved against it. */\n workdir?: string;\n /** Run in the background and return a task id immediately (collect with task_output, stop with task_kill). No timeout applies. */\n run_in_background?: boolean;\n } & Record;\n /** Inspect the live Cordis runtime in the current DSH process. Read-only. Sections: `services` (every provided ctx service and the plugin fiber that owns it), `plugins` (all live plugin fibers with their lifecycle states), `tools` (the model-facing tools currently registered, i.e. what you can call), `temporary` (only temporary Plugins created by cordis_mount: id, name, state, provided services, awaited services, and lifetime), `api` (method signatures AND argument/return type shapes for every LIVE service — read this before writing plugin code that calls a service), `events` (every harness event with its dispatch mode and exact signature — pick listener targets here). Temporary Plugins exist only in memory, remain active across later turns, and disappear after cordis_unmount, toolset unload, or DSH restart; they are not restored automatically. The `temporary` section is a subset of `plugins`. Omit `what` to get all six sections. With `what:\"api\"` or `what:\"events\"`, pass an exact `name` to narrow to one service/event and include its original source JSDoc. */\n cordis_inspect: {\n /** Limit the report to one section. Omit for all sections. */\n what?: \"services\" | \"plugins\" | \"tools\" | \"temporary\" | \"api\" | \"events\";\n /** Exact service key or event name whose original JSDoc to include; valid only with what:\"api\" or what:\"events\". */\n name?: string;\n } & Record;\n /** Mount a temporary Cordis Plugin in the current DSH process. This creates an in-memory runtime Plugin, not an installed or configured Plugin. It remains active across later turns until cordis_unmount, toolset unload, or DSH restart. It does not create files, install a package, change cordis.yml or personal/project config, survive restart, or automatically become permanent. To keep it, ask the Agent to implement a normal local, project, or repository Plugin through the regular development workflow. It may affect other sessions in the same process; the sandbox is not a security boundary, and injected services reach the real runtime. `code` runs now as the body of an async JavaScript function in an isolated sandbox and MUST `return` a plugin. Two forms: FUNCTION form `return (ctx) => { … }` — declares no inject, so it can register tools, listen to events, and provide services, but reaching ANY service (e.g. ctx.bash) throws; use it only when you need no services. OBJECT form `return { name?, inject: ['bash', 'llm', …], apply(ctx) { … } }` — declares dependencies, and cordis activates the plugin only after the services exist; PREFER this form. You may reach ONLY the services you list in inject: an undeclared service throws even if it exists, because an undeclared dependency would not be cleaned up if its provider is unmounted. BEFORE calling a service from your code, read cordis_inspect what:\"api\" — it lists method signatures AND the type shapes of their arguments/returns (do not guess a field's type; e.g. a bash run's stdout is an object, not a string). Inside `apply`, use the standard cordis API: `ctx.on(event, listener)` to observe events (see cordis_inspect what:\"events\"), or call `harness.registerTool(ctx, harness.defineTool({ name, description, parameters: { text: { type: 'string', required: true } }, output: { schema: { type: 'string' }, render(_args, value) { return [{ type: 'text', text: value }] } }, async execute(args) { return args.text } }))` to give yourself a new tool — it becomes callable on your NEXT step. Tool parameters: each key IS a property — { type: 'string'|'number'|'integer'|'boolean'|'null'|'object'|'array'|'json', required?: true, description?, enum?, const?, items?, properties? }; every direct DSL object declares additionalProperties: true|false, and oneOf: [schema, schema, ...] replaces type for an exact-one union. A raw JSON-Schema { type: 'object', properties, required?: […] } wrapper is also accepted with open-by-default objects. A tool's `execute` MUST return the lossless JSON value declared by `output.schema`; `output.render(args, value)` separately returns Native/model content blocks. Temporary Plugins can COMPOSE: one Plugin may `ctx.provide('name', value)` a service and another may declare `inject: ['name']` to consume it — the consumer stays pending until the provider exists and returns to pending when the provider is unmounted. Everything registered inside `apply` is cleaned up automatically by cordis_unmount. Sandbox globals: `console` (tagged `[cordis:]`, writes through to the harness terminal), `harness.defineTool`, `harness.registerTool`, `btoa`, `atob`, `TextEncoder`, `TextDecoder`. Node APIs are DISABLED — do filesystem/network/timer work through the cordis services, never Node built-ins: `require`, `setTimeout`/`setInterval`, and `fetch` throw redirect errors; `process` and `Buffer` are undefined. Instead use inject: ['fs'] + ctx.fs for files, inject: ['web'] + ctx.web for HTTP, inject: ['bash'] + ctx.bash for processes, and inject: ['timer'] + ctx.setTimeout/ctx.setInterval for timing (fiber effects, auto-cleaned when unmounted) — cordis_inspect what:\"api\" shows what THIS runtime provides. Write PLAIN JavaScript, not TypeScript (no `as`, no type annotations). Cautions: (1) waterfall events (e.g. tools/pre-execute) hand the listener a trailing `next` callback which MUST be called — returning without `next()` VETOES the call; prefer plain notification events unless you intend to intercept. (2) Never await something that only resolves after the current turn (your code runs INSIDE a tool call of that turn — it would deadlock). (3) Your `ctx` is a restricted façade: you can register tools, observe events, provide/consume services, and use timers, but framework internals (ctx.root, ctx.fiber, ctx.extend, ctx.plugin, …) are withheld. It is not a security boundary though — the services you inject (e.g. ctx.bash) reach the real runtime. */\n cordis_mount: {\n /** JavaScript body returning a temporary Plugin; evaluated now and saved nowhere. */\n code: string;\n } & Record;\n /** Unmount a current-process temporary Plugin created by cordis_mount. Waits for its tools, listeners, services, timers, and other owned effects to clean up completely. Only dyn-N temporary ids are accepted; this cannot remove Loader, configured, or installed Plugins. */\n cordis_unmount: {\n /** The temporary Plugin id returned by cordis_mount (for example \"dyn-1\"); valid only in this process and invalid after unmount or restart. */\n id: string;\n } & Record;\n /** Edit an existing UTF-8 text file by replacing literal text. */\n edit: {\n /** Path to edit, resolved by the filesystem backend. */\n file_path: string;\n /** Literal text to replace. Must match exactly. */\n old_string: string;\n /** Literal replacement text. Use an empty string to delete the match. */\n new_string: string;\n /** Replace all matches. Defaults to false; when false, old_string must appear exactly once. */\n replace_all?: boolean;\n } & Record;\n /** Run a foreground fresh-agent Ralph loop toward one immutable objective. Use only when the direct human explicitly asks for Ralph or fresh-agent iteration. Each round opens a new child with no parent conversation or prior child session; the shared workspace is long-term memory, and only a bounded structured report crosses rounds. The call returns when a worker reports completion or a concrete blocker, or at the round limit. Ordinary long-running same-session work belongs to goal tools. */\n ralph: {\n /** The immutable completion objective for every fresh Ralph round. */\n objective: string;\n /** Optional positive safe-integer round cap, bounded by the deployment ceiling. */\n maxRounds?: number;\n } & Record;\n /** Read a UTF-8 text file and return line-numbered content. */\n read: {\n /** Path to read, resolved by the filesystem backend. */\n file_path: string;\n /** 1-based first line to return. Defaults to 1. */\n offset?: number;\n /** Maximum number of lines to return. Defaults to 2000. */\n limit?: number;\n } & Record;\n /** Load the full instructions for an available skill. Call this with the exact skill name from the session skill catalog before acting on a task that names or clearly matches that skill. */\n skill: {\n /** The exact skill name from the available skills list. */\n name: string;\n } & Record;\n /** Delegate a self-contained task to a subagent (a separate agent that works in its own context) and return its final result. Use this to offload focused, independent work — research, a scoped implementation, an analysis — so it does not consume this conversation's context. The subagent runs to completion and you receive only its final answer, not its intermediate steps. Give it a complete, standalone prompt: it does not see this conversation. Set `run_in_background: true` to return a task id; collect with `task_output` and stop with `task_kill`. */\n subagent: {\n /** A short (3-5 word) description of the delegated task, for display. */\n description: string;\n /** The complete, self-contained task for the subagent. It does not share this conversation's context, so include everything it needs. */\n prompt: string;\n /** Run as a background task and return its id; collect with task_output or stop with task_kill. */\n run_in_background?: boolean;\n } & Record;\n /** Delegate a task to a subagent that inherits this conversation: a child agent seeded with all completed turns so far (it does not see the current in-flight turn), returning only its final result. Use this when the subtask builds on this conversation's context — a follow-up analysis, a review, a continuation — without consuming this conversation's context for the work itself. You receive only its final answer, not its intermediate steps. Set `run_in_background: true` to return a task id; collect with `task_output` and stop with `task_kill`. */\n subagent_fork: {\n /** A short (3-5 word) description of the delegated task, for display. */\n description: string;\n /** The task for the subagent. It already sees this conversation's completed turns, so build on them freely and state only what is new. */\n prompt: string;\n /** Run as a background task and return its id; collect with task_output or stop with task_kill. */\n run_in_background?: boolean;\n } & Record;\n /** Request cancellation of a running background task by task id. Returns immediately; the task settles as killed once its work actually stops. */\n task_kill: {\n /** Task id returned by the tool that started the background work. */\n task_id: string;\n /** Optional short reason, recorded in the log and forwarded to the task. */\n reason?: string;\n } & Record;\n /** List your background tasks (running and finished) with their ids, kinds, and statuses. */\n task_list: Record;\n /** Read a background task. Stream tasks return only output since the previous read; final-output tasks return their result after settlement. Every response ends with `[status: ...]`. Reads are non-blocking unless `wait: true`, which waits up to the configured cap. */\n task_output: {\n /** Task id returned by the tool that started the background work. */\n task_id: string;\n /** Block until the task reaches a terminal status or the timeout expires. A timed-out wait returns [status: running] and leaves the task alive. */\n wait?: boolean;\n /** Max wait in milliseconds (only meaningful with wait: true). Defaults to the configured wait timeout; capped by the configured maximum. */\n timeout_ms?: number;\n } & Record;\n /** Record and update a structured task list for the current work. Send the ENTIRE list every call — it REPLACES the previous list (there are no partial updates, no per-item edits). Use it to plan multi-step work and show progress: add one todo per concrete step before you start. Keep AT MOST ONE todo `in_progress` at a time; while work remains, exactly one active task should be `in_progress`. Mark a todo `completed` the moment it is done (do not batch completions), and allow no `in_progress` item only once all work is complete. Skip the list for trivial single-step tasks. Statuses: `pending` (not started), `in_progress` (being worked on now), `completed` (finished). */\n todo_write: {\n /** The COMPLETE task list, replacing any previous list. */\n todos: ({\n /** What the task is — a short imperative line. */\n content: string;\n /** pending (not started) | in_progress (now) | completed (done). */\n status: \"pending\" | \"in_progress\" | \"completed\";\n })[];\n } & Record;\n /** Run a JavaScript workflow script that orchestrates subagents at scale. Use this for work that fans out across many independent pieces — an audit over many files, a migration, multi-angle research, adversarial verification of findings — where you write the orchestration as a script instead of delegating turn by turn. The workflow's identity rides the `meta` parameter as JSON: required `name` (short kebab-case) and `description` strings, optional `whenToUse` string and `phases` array (`{title, detail?, provider?, model?}`). The `script` parameter is the plain JavaScript body ONLY (NOT TypeScript, and NO `export const meta` statement — meta is a parameter, not code), running with top-level await; end with `return ` — the value must be JSON-serializable and is this tool's result. Script-body hooks: - `agent(prompt, opts?): Promise` — run one subagent to completion. Without `opts.schema` it resolves to the child's final text; with `opts.schema` (an object-rooted JSON Schema using ONLY type/properties/required/additionalProperties/items/enum/const/oneOf — no pattern/format/numeric bounds) it resolves to the validated object. Resolves `null` when the child fails (filter with `.filter(Boolean)`). Other opts: `label` (display), `phase` (progress group), and independent `provider`/`model` LLM target overrides (either may be provided alone). Anything else (`effort`/`isolation`/`agentType`) is rejected loudly. - `pipeline(items, ...stages): Promise` — run each item through the stages independently with NO barrier between stages (prefer this for multi-stage work). Each stage receives `(prev, item, index)`. An ordinary stage throw drops that ITEM to `null` and skips its remaining stages. - `parallel(thunks): Promise` — run zero-argument functions concurrently and await ALL of them (a barrier; use only when a stage genuinely needs every prior result together). A throwing thunk resolves to `null`. - `phase(title)` — start a progress phase; `log(message)` — narrate progress; `args` — the tool call's `args` input, verbatim. Misused hooks (bad arguments, unknown options, unsupported schemas, tripped caps) throw errors that ALWAYS kill the script — they never dissolve into a per-item `null`. Constraints: concurrency and total-agent caps apply; no filesystem, network, timers, or Node.js APIs are provided — the agents do the work, the script only coordinates them. The run executes in the foreground: this call returns when the whole script finishes. */\n workflow: {\n /** The plain-JS workflow script body (top-level await allowed; NO `export const meta` statement; end with `return `). */\n script: string;\n /** The workflow identity block (plain JSON — never code). */\n meta: {\n /** Short kebab-case workflow name. */\n name: string;\n /** One-line description of what the workflow does. */\n description: string;\n /** Optional guidance on when this workflow applies. */\n whenToUse?: string;\n /** Optional phase declarations matched by phase() calls. */\n phases?: ({\n /** The phase title phase() calls match by exact string. */\n title: string;\n /** Optional one-line description of the phase. */\n detail?: string;\n /** Optional provider override this phase is expected to use. */\n provider?: string;\n /** Optional model override this phase is expected to use. */\n model?: string;\n } & Record)[];\n } & Record;\n /** Optional JSON input exposed to the script as the `args` global (wrap a bare list as a field, e.g. {\"files\": [...]}). */\n args?: Record;\n } & Record;\n /** Create or fully replace a UTF-8 text file. */\n write: {\n /** Path to write, resolved by the filesystem backend. */\n file_path: string;\n /** Full UTF-8 text content to write. */\n content: string;\n } & Record;\n}\n\ninterface ToolOutputMap {\n bash: {\n kind: \"background\";\n taskId: string;\n } | {\n kind: \"foreground\";\n exitCode: number | null;\n signal: string | null;\n timedOut: boolean;\n aborted: boolean;\n timeoutMs: number;\n stdout: {\n text: string;\n truncated: boolean;\n spillPath?: string;\n };\n stderr: {\n text: string;\n truncated: boolean;\n spillPath?: string;\n };\n sandbox?: {\n mode: string;\n denied: boolean;\n enforcement?: string;\n runnerFailed?: boolean;\n };\n };\n cordis_inspect: string;\n cordis_mount: {\n id: string;\n pluginName: string;\n state: \"pending\" | \"loading\" | \"active\" | \"failed\" | \"disposed\" | \"unloading\";\n provides: string[];\n waitingFor: string[];\n };\n cordis_unmount: {\n id: string;\n pluginName: string;\n };\n edit: {\n path: string;\n before: string;\n after: string;\n };\n ralph: {\n runId: string;\n agentsStarted: number;\n result: JsonValue;\n };\n read: {\n path: string;\n offset: number;\n lines: {\n number: number;\n text: string;\n }[];\n totalLines: number;\n };\n skill: {\n name: string;\n provider: string;\n resourceBase?: {\n kind: \"directory\";\n path: string;\n } | {\n kind: \"url\";\n url: string;\n } | {\n kind: \"opaque\";\n description: string;\n };\n content: string;\n };\n subagent: {\n kind: \"background\";\n taskId: string;\n } | {\n kind: \"foreground\";\n runId: string;\n output: JsonValue[];\n };\n subagent_fork: {\n kind: \"background\";\n taskId: string;\n } | {\n kind: \"foreground\";\n runId: string;\n output: JsonValue[];\n };\n task_kill: {\n outcome: \"cancellation-requested\" | \"already-finished\";\n task: {\n id: string;\n kind: string;\n label: string;\n status: \"running\" | \"stopping\" | \"completed\" | \"killed\" | \"failed\";\n detail?: string;\n startedAt: number;\n finishedAt?: number;\n };\n };\n task_list: ({\n id: string;\n kind: string;\n label: string;\n status: \"running\" | \"stopping\" | \"completed\" | \"killed\" | \"failed\";\n detail?: string;\n startedAt: number;\n finishedAt?: number;\n })[];\n task_output: {\n text: string;\n task: {\n id: string;\n kind: string;\n label: string;\n status: \"running\" | \"stopping\" | \"completed\" | \"killed\" | \"failed\";\n detail?: string;\n startedAt: number;\n finishedAt?: number;\n };\n };\n todo_write: {\n todos: ({\n content: string;\n status: \"pending\" | \"in_progress\" | \"completed\";\n })[];\n counts: {\n pending: number;\n inProgress: number;\n completed: number;\n };\n };\n workflow: {\n runId: string;\n agentsStarted: number;\n result: JsonValue;\n };\n write: {\n path: string;\n operation: \"create\" | \"update\";\n before: string | null;\n after: string;\n };\n}\n\ntype ToolName = keyof ToolOutputMap\n\ndeclare class ToolCallError extends Error {\n readonly name: \"ToolCallError\";\n readonly toolName: ToolName;\n}\n\ndeclare const tools: {\n [K in ToolName]: (args: ToolArgsMap[K]) => Promise;\n}\n```","tools":[{"name":"bash","description":"Execute a bash command (`bash -c`) and return its stdout/stderr. Each call runs in a fresh shell: no state (cwd, variables, functions) persists between calls — pass `workdir` instead of using `cd`. Non-zero exits are reported as `[exit code: N]`. Current harness environment facts are exposed through managed `$DSH_*` variables; inspect them when needed. Commands may run under a file sandbox; a blocked file operation is reported as `[sandbox: file access denied under mode]` — a policy denial, not a bug in the command; do not retry another way. Long output is truncated to its tail; the full output is saved to a file whose path is reported when available. Set `run_in_background: true` for long-running commands: the call returns a task id immediately; read its output with `task_output` and stop it with `task_kill`.","parameters":{"type":"object","properties":{"command":{"type":"string","description":"The bash command to execute."},"description":{"type":"string","description":"Clear, concise description of what this command does in active voice, 5-10 words (shown in the UI). Examples: \"ls\" → \"List files in current directory\"; \"git status\" → \"Show working tree status\"; \"npm install\" → \"Install package dependencies\"."},"timeoutMs":{"type":"number","description":"Timeout in milliseconds. The executor applies its configured default and cap, and kills the command on expiry."},"workdir":{"type":"string","description":"Working directory for this command. Defaults to the session workspace; a relative path is resolved against it."},"run_in_background":{"type":"boolean","description":"Run in the background and return a task id immediately (collect with task_output, stop with task_kill). No timeout applies."}},"required":["command","description"]}},{"name":"cordis_inspect","description":"Inspect the live Cordis runtime in the current DSH process. Read-only. Sections: `services` (every provided ctx service and the plugin fiber that owns it), `plugins` (all live plugin fibers with their lifecycle states), `tools` (the model-facing tools currently registered, i.e. what you can call), `temporary` (only temporary Plugins created by cordis_mount: id, name, state, provided services, awaited services, and lifetime), `api` (method signatures AND argument/return type shapes for every LIVE service — read this before writing plugin code that calls a service), `events` (every harness event with its dispatch mode and exact signature — pick listener targets here). Temporary Plugins exist only in memory, remain active across later turns, and disappear after cordis_unmount, toolset unload, or DSH restart; they are not restored automatically. The `temporary` section is a subset of `plugins`. Omit `what` to get all six sections. With `what:\"api\"` or `what:\"events\"`, pass an exact `name` to narrow to one service/event and include its original source JSDoc.","parameters":{"type":"object","properties":{"what":{"type":"string","description":"Limit the report to one section. Omit for all sections.","enum":["services","plugins","tools","temporary","api","events"]},"name":{"type":"string","description":"Exact service key or event name whose original JSDoc to include; valid only with what:\"api\" or what:\"events\"."}}}},{"name":"cordis_mount","description":"Mount a temporary Cordis Plugin in the current DSH process. This creates an in-memory runtime Plugin, not an installed or configured Plugin. It remains active across later turns until cordis_unmount, toolset unload, or DSH restart. It does not create files, install a package, change cordis.yml or personal/project config, survive restart, or automatically become permanent. To keep it, ask the Agent to implement a normal local, project, or repository Plugin through the regular development workflow. It may affect other sessions in the same process; the sandbox is not a security boundary, and injected services reach the real runtime. `code` runs now as the body of an async JavaScript function in an isolated sandbox and MUST `return` a plugin. Two forms: FUNCTION form `return (ctx) => { … }` — declares no inject, so it can register tools, listen to events, and provide services, but reaching ANY service (e.g. ctx.bash) throws; use it only when you need no services. OBJECT form `return { name?, inject: ['bash', 'llm', …], apply(ctx) { … } }` — declares dependencies, and cordis activates the plugin only after the services exist; PREFER this form. You may reach ONLY the services you list in inject: an undeclared service throws even if it exists, because an undeclared dependency would not be cleaned up if its provider is unmounted. BEFORE calling a service from your code, read cordis_inspect what:\"api\" — it lists method signatures AND the type shapes of their arguments/returns (do not guess a field's type; e.g. a bash run's stdout is an object, not a string). Inside `apply`, use the standard cordis API: `ctx.on(event, listener)` to observe events (see cordis_inspect what:\"events\"), or call `harness.registerTool(ctx, harness.defineTool({ name, description, parameters: { text: { type: 'string', required: true } }, output: { schema: { type: 'string' }, render(_args, value) { return [{ type: 'text', text: value }] } }, async execute(args) { return args.text } }))` to give yourself a new tool — it becomes callable on your NEXT step. Tool parameters: each key IS a property — { type: 'string'|'number'|'integer'|'boolean'|'null'|'object'|'array'|'json', required?: true, description?, enum?, const?, items?, properties? }; every direct DSL object declares additionalProperties: true|false, and oneOf: [schema, schema, ...] replaces type for an exact-one union. A raw JSON-Schema { type: 'object', properties, required?: […] } wrapper is also accepted with open-by-default objects. A tool's `execute` MUST return the lossless JSON value declared by `output.schema`; `output.render(args, value)` separately returns Native/model content blocks. Temporary Plugins can COMPOSE: one Plugin may `ctx.provide('name', value)` a service and another may declare `inject: ['name']` to consume it — the consumer stays pending until the provider exists and returns to pending when the provider is unmounted. Everything registered inside `apply` is cleaned up automatically by cordis_unmount. Sandbox globals: `console` (tagged `[cordis:]`, writes through to the harness terminal), `harness.defineTool`, `harness.registerTool`, `btoa`, `atob`, `TextEncoder`, `TextDecoder`. Node APIs are DISABLED — do filesystem/network/timer work through the cordis services, never Node built-ins: `require`, `setTimeout`/`setInterval`, and `fetch` throw redirect errors; `process` and `Buffer` are undefined. Instead use inject: ['fs'] + ctx.fs for files, inject: ['web'] + ctx.web for HTTP, inject: ['bash'] + ctx.bash for processes, and inject: ['timer'] + ctx.setTimeout/ctx.setInterval for timing (fiber effects, auto-cleaned when unmounted) — cordis_inspect what:\"api\" shows what THIS runtime provides. Write PLAIN JavaScript, not TypeScript (no `as`, no type annotations). Cautions: (1) waterfall events (e.g. tools/pre-execute) hand the listener a trailing `next` callback which MUST be called — returning without `next()` VETOES the call; prefer plain notification events unless you intend to intercept. (2) Never await something that only resolves after the current turn (your code runs INSIDE a tool call of that turn — it would deadlock). (3) Your `ctx` is a restricted façade: you can register tools, observe events, provide/consume services, and use timers, but framework internals (ctx.root, ctx.fiber, ctx.extend, ctx.plugin, …) are withheld. It is not a security boundary though — the services you inject (e.g. ctx.bash) reach the real runtime.","parameters":{"type":"object","properties":{"code":{"type":"string","description":"JavaScript body returning a temporary Plugin; evaluated now and saved nowhere."}},"required":["code"]}},{"name":"cordis_unmount","description":"Unmount a current-process temporary Plugin created by cordis_mount. Waits for its tools, listeners, services, timers, and other owned effects to clean up completely. Only dyn-N temporary ids are accepted; this cannot remove Loader, configured, or installed Plugins.","parameters":{"type":"object","properties":{"id":{"type":"string","description":"The temporary Plugin id returned by cordis_mount (for example \"dyn-1\"); valid only in this process and invalid after unmount or restart."}},"required":["id"]}},{"name":"edit","description":"Edit an existing UTF-8 text file by replacing literal text.","parameters":{"type":"object","properties":{"file_path":{"type":"string","description":"Path to edit, resolved by the filesystem backend."},"old_string":{"type":"string","description":"Literal text to replace. Must match exactly."},"new_string":{"type":"string","description":"Literal replacement text. Use an empty string to delete the match."},"replace_all":{"type":"boolean","description":"Replace all matches. Defaults to false; when false, old_string must appear exactly once."}},"required":["file_path","old_string","new_string"]}},{"name":"ralph","description":"Run a foreground fresh-agent Ralph loop toward one immutable objective. Use only when the direct human explicitly asks for Ralph or fresh-agent iteration. Each round opens a new child with no parent conversation or prior child session; the shared workspace is long-term memory, and only a bounded structured report crosses rounds. The call returns when a worker reports completion or a concrete blocker, or at the round limit. Ordinary long-running same-session work belongs to goal tools.","parameters":{"type":"object","properties":{"objective":{"type":"string","description":"The immutable completion objective for every fresh Ralph round."},"maxRounds":{"type":"number","description":"Optional positive safe-integer round cap, bounded by the deployment ceiling."}},"required":["objective"]}},{"name":"read","description":"Read a UTF-8 text file and return line-numbered content.","parameters":{"type":"object","properties":{"file_path":{"type":"string","description":"Path to read, resolved by the filesystem backend."},"offset":{"type":"number","description":"1-based first line to return. Defaults to 1."},"limit":{"type":"number","description":"Maximum number of lines to return. Defaults to 2000."}},"required":["file_path"]}},{"name":"run_code","description":"Execute a TypeScript program against the available tools. Write the BODY of an async function (erasable syntax only; top-level `await` and `return` work) and call tools as `await tools.name(args)` per the declarations in the system prompt. Only what you print or return comes back — curate it.","parameters":{"type":"object","properties":{"code":{"type":"string","description":"The program: the body of an async TypeScript function."},"description":{"type":"string","description":"Clear, concise description of what this program does in active voice, 5-10 words (shown in the UI). Examples: \"Count TODO markers across packages\"; \"Read failing test and its fixture\"; \"Rename config key in every cordis.yml\"."}},"required":["code","description"]}},{"name":"skill","description":"Load the full instructions for an available skill. Call this with the exact skill name from the session skill catalog before acting on a task that names or clearly matches that skill.","parameters":{"type":"object","properties":{"name":{"type":"string","description":"The exact skill name from the available skills list."}},"required":["name"]}},{"name":"subagent","description":"Delegate a self-contained task to a subagent (a separate agent that works in its own context) and return its final result. Use this to offload focused, independent work — research, a scoped implementation, an analysis — so it does not consume this conversation's context. The subagent runs to completion and you receive only its final answer, not its intermediate steps. Give it a complete, standalone prompt: it does not see this conversation. Set `run_in_background: true` to return a task id; collect with `task_output` and stop with `task_kill`.","parameters":{"type":"object","properties":{"description":{"type":"string","description":"A short (3-5 word) description of the delegated task, for display."},"prompt":{"type":"string","description":"The complete, self-contained task for the subagent. It does not share this conversation's context, so include everything it needs."},"run_in_background":{"type":"boolean","description":"Run as a background task and return its id; collect with task_output or stop with task_kill."}},"required":["description","prompt"]}},{"name":"subagent_fork","description":"Delegate a task to a subagent that inherits this conversation: a child agent seeded with all completed turns so far (it does not see the current in-flight turn), returning only its final result. Use this when the subtask builds on this conversation's context — a follow-up analysis, a review, a continuation — without consuming this conversation's context for the work itself. You receive only its final answer, not its intermediate steps. Set `run_in_background: true` to return a task id; collect with `task_output` and stop with `task_kill`.","parameters":{"type":"object","properties":{"description":{"type":"string","description":"A short (3-5 word) description of the delegated task, for display."},"prompt":{"type":"string","description":"The task for the subagent. It already sees this conversation's completed turns, so build on them freely and state only what is new."},"run_in_background":{"type":"boolean","description":"Run as a background task and return its id; collect with task_output or stop with task_kill."}},"required":["description","prompt"]}},{"name":"task_kill","description":"Request cancellation of a running background task by task id. Returns immediately; the task settles as killed once its work actually stops.","parameters":{"type":"object","properties":{"task_id":{"type":"string","description":"Task id returned by the tool that started the background work."},"reason":{"type":"string","description":"Optional short reason, recorded in the log and forwarded to the task."}},"required":["task_id"]}},{"name":"task_list","description":"List your background tasks (running and finished) with their ids, kinds, and statuses.","parameters":{"type":"object","properties":{}}},{"name":"task_output","description":"Read a background task. Stream tasks return only output since the previous read; final-output tasks return their result after settlement. Every response ends with `[status: ...]`. Reads are non-blocking unless `wait: true`, which waits up to the configured cap.","parameters":{"type":"object","properties":{"task_id":{"type":"string","description":"Task id returned by the tool that started the background work."},"wait":{"type":"boolean","description":"Block until the task reaches a terminal status or the timeout expires. A timed-out wait returns [status: running] and leaves the task alive."},"timeout_ms":{"type":"number","description":"Max wait in milliseconds (only meaningful with wait: true). Defaults to the configured wait timeout; capped by the configured maximum."}},"required":["task_id"]}},{"name":"todo_write","description":"Record and update a structured task list for the current work. Send the ENTIRE list every call — it REPLACES the previous list (there are no partial updates, no per-item edits). Use it to plan multi-step work and show progress: add one todo per concrete step before you start. Keep AT MOST ONE todo `in_progress` at a time; while work remains, exactly one active task should be `in_progress`. Mark a todo `completed` the moment it is done (do not batch completions), and allow no `in_progress` item only once all work is complete. Skip the list for trivial single-step tasks. Statuses: `pending` (not started), `in_progress` (being worked on now), `completed` (finished).","parameters":{"type":"object","properties":{"todos":{"type":"array","description":"The COMPLETE task list, replacing any previous list.","items":{"type":"object","additionalProperties":false,"properties":{"content":{"type":"string","description":"What the task is — a short imperative line."},"status":{"type":"string","description":"pending (not started) | in_progress (now) | completed (done).","enum":["pending","in_progress","completed"]}},"required":["content","status"]}}},"required":["todos"]}},{"name":"workflow","description":"Run a JavaScript workflow script that orchestrates subagents at scale. Use this for work that fans out across many independent pieces — an audit over many files, a migration, multi-angle research, adversarial verification of findings — where you write the orchestration as a script instead of delegating turn by turn.\n\nThe workflow's identity rides the `meta` parameter as JSON: required `name` (short kebab-case) and `description` strings, optional `whenToUse` string and `phases` array (`{title, detail?, provider?, model?}`). The `script` parameter is the plain JavaScript body ONLY (NOT TypeScript, and NO `export const meta` statement — meta is a parameter, not code), running with top-level await; end with `return ` — the value must be JSON-serializable and is this tool's result.\n\nScript-body hooks:\n- `agent(prompt, opts?): Promise` — run one subagent to completion. Without `opts.schema` it resolves to the child's final text; with `opts.schema` (an object-rooted JSON Schema using ONLY type/properties/required/additionalProperties/items/enum/const/oneOf — no pattern/format/numeric bounds) it resolves to the validated object. Resolves `null` when the child fails (filter with `.filter(Boolean)`). Other opts: `label` (display), `phase` (progress group), and independent `provider`/`model` LLM target overrides (either may be provided alone). Anything else (`effort`/`isolation`/`agentType`) is rejected loudly.\n- `pipeline(items, ...stages): Promise` — run each item through the stages independently with NO barrier between stages (prefer this for multi-stage work). Each stage receives `(prev, item, index)`. An ordinary stage throw drops that ITEM to `null` and skips its remaining stages.\n- `parallel(thunks): Promise` — run zero-argument functions concurrently and await ALL of them (a barrier; use only when a stage genuinely needs every prior result together). A throwing thunk resolves to `null`.\n- `phase(title)` — start a progress phase; `log(message)` — narrate progress; `args` — the tool call's `args` input, verbatim.\n\nMisused hooks (bad arguments, unknown options, unsupported schemas, tripped caps) throw errors that ALWAYS kill the script — they never dissolve into a per-item `null`.\n\nConstraints: concurrency and total-agent caps apply; no filesystem, network, timers, or Node.js APIs are provided — the agents do the work, the script only coordinates them. The run executes in the foreground: this call returns when the whole script finishes.","parameters":{"type":"object","properties":{"script":{"type":"string","description":"The plain-JS workflow script body (top-level await allowed; NO `export const meta` statement; end with `return `)."},"meta":{"type":"object","description":"The workflow identity block (plain JSON — never code).","additionalProperties":true,"properties":{"name":{"type":"string","description":"Short kebab-case workflow name."},"description":{"type":"string","description":"One-line description of what the workflow does."},"whenToUse":{"type":"string","description":"Optional guidance on when this workflow applies."},"phases":{"type":"array","description":"Optional phase declarations matched by phase() calls.","items":{"type":"object","additionalProperties":true,"properties":{"title":{"type":"string","description":"The phase title phase() calls match by exact string."},"detail":{"type":"string","description":"Optional one-line description of the phase."},"provider":{"type":"string","description":"Optional provider override this phase is expected to use."},"model":{"type":"string","description":"Optional model override this phase is expected to use."}},"required":["title"]}}},"required":["name","description"]},"args":{"type":"object","description":"Optional JSON input exposed to the script as the `args` global (wrap a bare list as a field, e.g. {\"files\": [...]}).","additionalProperties":true}},"required":["script","meta"]}},{"name":"write","description":"Create or fully replace a UTF-8 text file.","parameters":{"type":"object","properties":{"file_path":{"type":"string","description":"Path to write, resolved by the filesystem backend."},"content":{"type":"string","description":"Full UTF-8 text content to write."}},"required":["file_path","content"]}}]},"reason":"initial"}} +{"type":"request/header","seq":4,"time":1783957884564,"data":{"header":{"config":{"provider":"deepseek","model":"deepseek-v4-flash"},"system":"You are an AI agent powered by the DeepSeek Harness SDK.\n\nYou are headless-agent, a coding assistant powered by the deepseek-v4-flash model. Your working directory is {{cwd}}.\n\nVerify your work by running the code or tests. Keep answers brief and factual.\n\n\nUse the read tool — not shell commands like cat — to inspect text files. Results include line numbers. Use offset and limit to continue reading large files.\n\nUse the write tool to create files or completely replace file contents. Existing files are overwritten, so read an existing file first (the default fs-policy requires it) and prefer edit for targeted changes.\n\nUse the edit tool for targeted changes to existing UTF-8 text files. It replaces literal old_string with new_string; by default old_string must appear exactly once. If old_string appears multiple times, provide a more specific old_string or set replace_all to true. Read the file first (the default fs-policy requires it), unless you just created or edited it in this session.\n\nCheck the [exit code: N] marker on every bash result; investigate failures before moving on.\n\nTrack every background task id you start. You are notified in-session when a task finishes — do not busy-poll or sleep on one; keep working on independent steps and do not duplicate a running task's work. Before giving a final answer, collect every still-relevant task with task_output (set wait: true only when you are genuinely blocked on it), and task_kill tasks that stopped mattering.\n\nUse the workflow tool ONLY when the user explicitly asks for a workflow or for large multi-agent orchestration: you write a JavaScript script (the tool description documents the exact format) that fans work out across many subagents with phases and structured results. For one or two delegations, prefer plain subagent calls.\n\nUse the ralph tool ONLY when the direct human explicitly asks for a Ralph loop or fresh-agent iterative execution. Each Ralph round starts a fresh child with no conversation seed and uses the shared workspace as durable memory. Completion and blockers are worker reports, not independent evaluation. Use same-session goal tools for ordinary long-running objectives, and plain subagents or workflows for bounded delegation and fan-out.\n\n## Writing code for run_code\n\nPass `run_code` the body of an async TypeScript function (erasable syntax only — no `enum` or namespaces; type annotations are advisory, the code runs type-stripped). Inside the program:\n\n- Call tools as `await tools.name(args)` — quoted access for exotic names: `tools[\"my-tool\"](args)`. Every call resolves to the tool's typed canonical JSON value. Tool arguments must be lossless JSON.\n- A FAILED tool call rejects with `ToolCallError`, whose `toolName` identifies the failed tool and whose `message` is human-readable — `try/catch` it to handle and continue.\n- Independent read-only calls MAY overlap under `Promise.all` (safe calls run concurrently; mutating calls run alone, in submission order). Sequence dependent work with `await`.\n- Emit results with `return` and/or `console.log(...)`. ONLY what you print or return comes back to you — intermediate tool results never enter the conversation, so extract just what you need.\n\nThe available tools:\n\n```ts\ntype JsonValue = null | boolean | number | string | JsonValue[] | { [key: string]: JsonValue }\n\ninterface ToolArgsMap {\n /** Execute a bash command (`bash -c`) and return its stdout/stderr. Each call runs in a fresh shell: no state (cwd, variables, functions) persists between calls — pass `workdir` instead of using `cd`. Non-zero exits are reported as `[exit code: N]`. Current harness environment facts are exposed through managed `$DSH_*` variables; inspect them when needed. Commands may run under a file sandbox; a blocked file operation is reported as `[sandbox: file access denied under mode]` — a policy denial, not a bug in the command; do not retry another way. Long output is truncated to its tail; the full output is saved to a file whose path is reported when available. Set `run_in_background: true` for long-running commands: the call returns a task id immediately; read its output with `task_output` and stop it with `task_kill`. */\n bash: {\n /** The bash command to execute. */\n command: string;\n /** Clear, concise description of what this command does in active voice, 5-10 words (shown in the UI). Examples: \"ls\" → \"List files in current directory\"; \"git status\" → \"Show working tree status\"; \"npm install\" → \"Install package dependencies\". */\n description: string;\n /** Timeout in milliseconds. The executor applies its configured default and cap, and kills the command on expiry. */\n timeoutMs?: number;\n /** Working directory for this command. Defaults to the session workspace; a relative path is resolved against it. */\n workdir?: string;\n /** Run in the background and return a task id immediately (collect with task_output, stop with task_kill). No timeout applies. */\n run_in_background?: boolean;\n } & Record;\n /** Inspect the live Cordis runtime in the current DSH process. Read-only. Sections: `services` (every provided ctx service and the plugin fiber that owns it), `plugins` (all live plugin fibers with their lifecycle states), `tools` (the model-facing tools currently registered, i.e. what you can call), `temporary` (only temporary Plugins created by cordis_mount: id, name, state, provided services, awaited services, and lifetime), `api` (method signatures AND argument/return type shapes for every LIVE service — read this before writing plugin code that calls a service), `events` (every harness event with its dispatch mode and exact signature — pick listener targets here). Temporary Plugins exist only in memory, remain active across later turns, and disappear after cordis_unmount, toolset unload, or DSH restart; they are not restored automatically. The `temporary` section is a subset of `plugins`. Omit `what` to get all six sections. With `what:\"api\"` or `what:\"events\"`, pass an exact `name` to narrow to one service/event and include its original source JSDoc. */\n cordis_inspect: {\n /** Limit the report to one section. Omit for all sections. */\n what?: \"services\" | \"plugins\" | \"tools\" | \"temporary\" | \"api\" | \"events\";\n /** Exact service key or event name whose original JSDoc to include; valid only with what:\"api\" or what:\"events\". */\n name?: string;\n } & Record;\n /** Mount a temporary Cordis Plugin in the current DSH process. This creates an in-memory runtime Plugin, not an installed or configured Plugin. It remains active across later turns until cordis_unmount, toolset unload, or DSH restart. It does not create files, install a package, change cordis.yml or personal/project config, survive restart, or automatically become permanent. To keep it, ask the Agent to implement a normal local, project, or repository Plugin through the regular development workflow. It may affect other sessions in the same process; the sandbox is not a security boundary, and injected services reach the real runtime. `code` runs now as the body of an async JavaScript function in an isolated sandbox and MUST `return` a plugin. Two forms: FUNCTION form `return (ctx) => { … }` — declares no inject, so it can register tools, listen to events, and provide services, but reaching ANY service (e.g. ctx.bash) throws; use it only when you need no services. OBJECT form `return { name?, inject: ['bash', 'llm', …], apply(ctx) { … } }` — declares dependencies, and cordis activates the plugin only after the services exist; PREFER this form. You may reach ONLY the services you list in inject: an undeclared service throws even if it exists, because an undeclared dependency would not be cleaned up if its provider is unmounted. BEFORE calling a service from your code, read cordis_inspect what:\"api\" — it lists method signatures AND the type shapes of their arguments/returns (do not guess a field's type; e.g. a bash run's stdout is an object, not a string). Inside `apply`, use the standard cordis API: `ctx.on(event, listener)` to observe events (see cordis_inspect what:\"events\"), or call `harness.registerTool(ctx, harness.defineTool({ name, description, parameters: { text: { type: 'string', required: true } }, output: { schema: { type: 'string' }, render(_args, value) { return [{ type: 'text', text: value }] } }, async execute(args) { return args.text } }))` to give yourself a new tool — it becomes callable on your NEXT step. Tool parameters: each key IS a property — { type: 'string'|'number'|'integer'|'boolean'|'null'|'object'|'array'|'json', required?: true, description?, enum?, const?, items?, properties? }; every direct DSL object declares additionalProperties: true|false, and oneOf: [schema, schema, ...] replaces type for an exact-one union. A raw JSON-Schema { type: 'object', properties, required?: […] } wrapper is also accepted with open-by-default objects. A tool's `execute` MUST return the lossless JSON value declared by `output.schema`; `output.render(args, value)` separately returns Native/model content blocks. Temporary Plugins can COMPOSE: one Plugin may `ctx.provide('name', value)` a service and another may declare `inject: ['name']` to consume it — the consumer stays pending until the provider exists and returns to pending when the provider is unmounted. Everything registered inside `apply` is cleaned up automatically by cordis_unmount. Sandbox globals: `console` (tagged `[cordis:]`, writes through to the harness terminal), `harness.defineTool`, `harness.registerTool`, `btoa`, `atob`, `TextEncoder`, `TextDecoder`. Node APIs are DISABLED — do filesystem/network/timer work through the cordis services, never Node built-ins: `require`, `setTimeout`/`setInterval`, and `fetch` throw redirect errors; `process` and `Buffer` are undefined. Instead use inject: ['fs'] + ctx.fs for files, inject: ['web'] + ctx.web for HTTP, inject: ['bash'] + ctx.bash for processes, and inject: ['timer'] + ctx.setTimeout/ctx.setInterval for timing (fiber effects, auto-cleaned when unmounted) — cordis_inspect what:\"api\" shows what THIS runtime provides. Write PLAIN JavaScript, not TypeScript (no `as`, no type annotations). Cautions: (1) waterfall events (e.g. tools/pre-execute) hand the listener a trailing `next` callback which MUST be called — returning without `next()` VETOES the call; prefer plain notification events unless you intend to intercept. (2) Never await something that only resolves after the current turn (your code runs INSIDE a tool call of that turn — it would deadlock). (3) Your `ctx` is a restricted façade: you can register tools, observe events, provide/consume services, and use timers, but framework internals (ctx.root, ctx.fiber, ctx.extend, ctx.plugin, …) are withheld. It is not a security boundary though — the services you inject (e.g. ctx.bash) reach the real runtime. */\n cordis_mount: {\n /** JavaScript body returning a temporary Plugin; evaluated now and saved nowhere. */\n code: string;\n } & Record;\n /** Unmount a current-process temporary Plugin created by cordis_mount. Waits for its tools, listeners, services, timers, and other owned effects to clean up completely. Only dyn-N temporary ids are accepted; this cannot remove Loader, configured, or installed Plugins. */\n cordis_unmount: {\n /** The temporary Plugin id returned by cordis_mount (for example \"dyn-1\"); valid only in this process and invalid after unmount or restart. */\n id: string;\n } & Record;\n /** Edit an existing UTF-8 text file by replacing literal text. */\n edit: {\n /** Path to edit, resolved by the filesystem backend. */\n file_path: string;\n /** Literal text to replace. Must match exactly. */\n old_string: string;\n /** Literal replacement text. Use an empty string to delete the match. */\n new_string: string;\n /** Replace all matches. Defaults to false; when false, old_string must appear exactly once. */\n replace_all?: boolean;\n } & Record;\n /** Run a foreground fresh-agent Ralph loop toward one immutable objective. Use only when the direct human explicitly asks for Ralph or fresh-agent iteration. Each round opens a new child with no parent conversation or prior child session; the shared workspace is long-term memory, and only a bounded structured report crosses rounds. The call returns when a worker reports completion or a concrete blocker, or at the round limit. Ordinary long-running same-session work belongs to goal tools. */\n ralph: {\n /** The immutable completion objective for every fresh Ralph round. */\n objective: string;\n /** Optional positive safe-integer round cap, bounded by the deployment ceiling. */\n maxRounds?: number;\n } & Record;\n /** Read a UTF-8 text file and return line-numbered content. */\n read: {\n /** Path to read, resolved by the filesystem backend. */\n file_path: string;\n /** 1-based first line to return. Defaults to 1. */\n offset?: number;\n /** Maximum number of lines to return. Defaults to 2000. */\n limit?: number;\n } & Record;\n /** Load the full instructions for an available skill. Call this with the exact skill name from the session skill catalog before acting on a task that names or clearly matches that skill. */\n skill: {\n /** The exact skill name from the available skills list. */\n name: string;\n } & Record;\n /** Delegate a self-contained task to a subagent (a separate agent that works in its own context) and return its final result. Use this to offload focused, independent work — research, a scoped implementation, an analysis — so it does not consume this conversation's context. The subagent runs to completion and you receive only its final answer, not its intermediate steps. Give it a complete, standalone prompt: it does not see this conversation. Set `run_in_background: true` to return a task id; collect with `task_output` and stop with `task_kill`. */\n subagent: {\n /** A short (3-5 word) description of the delegated task, for display. */\n description: string;\n /** The complete, self-contained task for the subagent. It does not share this conversation's context, so include everything it needs. */\n prompt: string;\n /** Run as a background task and return its id; collect with task_output or stop with task_kill. */\n run_in_background?: boolean;\n } & Record;\n /** Delegate a task to a subagent that inherits this conversation: a child agent seeded with all completed turns so far (it does not see the current in-flight turn), returning only its final result. Use this when the subtask builds on this conversation's context — a follow-up analysis, a review, a continuation — without consuming this conversation's context for the work itself. You receive only its final answer, not its intermediate steps. Set `run_in_background: true` to return a task id; collect with `task_output` and stop with `task_kill`. */\n subagent_fork: {\n /** A short (3-5 word) description of the delegated task, for display. */\n description: string;\n /** The task for the subagent. It already sees this conversation's completed turns, so build on them freely and state only what is new. */\n prompt: string;\n /** Run as a background task and return its id; collect with task_output or stop with task_kill. */\n run_in_background?: boolean;\n } & Record;\n /** Request cancellation of a running background task by task id. Returns immediately; the task settles as killed once its work actually stops. */\n task_kill: {\n /** Task id returned by the tool that started the background work. */\n task_id: string;\n /** Optional short reason, recorded in the log and forwarded to the task. */\n reason?: string;\n } & Record;\n /** List your background tasks (running and finished) with their ids, kinds, and statuses. */\n task_list: Record;\n /** Read a background task. Stream tasks return only output since the previous read; final-output tasks return their result after settlement. Every response ends with `[status: ...]`. Reads are non-blocking unless `wait: true`, which waits up to the configured cap. */\n task_output: {\n /** Task id returned by the tool that started the background work. */\n task_id: string;\n /** Block until the task reaches a terminal status or the timeout expires. A timed-out wait returns [status: running] and leaves the task alive. */\n wait?: boolean;\n /** Max wait in milliseconds (only meaningful with wait: true). Defaults to the configured wait timeout; capped by the configured maximum. */\n timeout_ms?: number;\n } & Record;\n /** Record and update a structured task list for the current work. Send the ENTIRE list every call — it REPLACES the previous list (there are no partial updates, no per-item edits). Use it to plan multi-step work and show progress: add one todo per concrete step before you start. Keep AT MOST ONE todo `in_progress` at a time; while work remains, exactly one active task should be `in_progress`. Mark a todo `completed` the moment it is done (do not batch completions), and allow no `in_progress` item only once all work is complete. Skip the list for trivial single-step tasks. Statuses: `pending` (not started), `in_progress` (being worked on now), `completed` (finished). */\n todo_write: {\n /** The COMPLETE task list, replacing any previous list. */\n todos: ({\n /** What the task is — a short imperative line. */\n content: string;\n /** pending (not started) | in_progress (now) | completed (done). */\n status: \"pending\" | \"in_progress\" | \"completed\";\n })[];\n } & Record;\n /** Run a JavaScript workflow script that orchestrates subagents at scale. Use this for work that fans out across many independent pieces — an audit over many files, a migration, multi-angle research, adversarial verification of findings — where you write the orchestration as a script instead of delegating turn by turn. The workflow's identity rides the `meta` parameter as JSON: required `name` (short kebab-case) and `description` strings, optional `whenToUse` string and `phases` array (`{title, detail?, provider?, model?}`). The `script` parameter is the plain JavaScript body ONLY (NOT TypeScript, and NO `export const meta` statement — meta is a parameter, not code), running with top-level await; end with `return ` — the value must be JSON-serializable and is this tool's result. Script-body hooks: - `agent(prompt, opts?): Promise` — run one subagent to completion. Without `opts.schema` it resolves to the child's final text; with `opts.schema` (an object-rooted JSON Schema using ONLY type/properties/required/additionalProperties/items/enum/const/oneOf — no pattern/format/numeric bounds) it resolves to the validated object. Resolves `null` when the child fails (filter with `.filter(Boolean)`). Other opts: `label` (display), `phase` (progress group), and independent `provider`/`model` LLM target overrides (either may be provided alone). Anything else (`effort`/`isolation`/`agentType`) is rejected loudly. - `pipeline(items, ...stages): Promise` — run each item through the stages independently with NO barrier between stages (prefer this for multi-stage work). Each stage receives `(prev, item, index)`. An ordinary stage throw drops that ITEM to `null` and skips its remaining stages. - `parallel(thunks): Promise` — run zero-argument functions concurrently and await ALL of them (a barrier; use only when a stage genuinely needs every prior result together). A throwing thunk resolves to `null`. - `phase(title)` — start a progress phase; `log(message)` — narrate progress; `args` — the tool call's `args` input, verbatim. Misused hooks (bad arguments, unknown options, unsupported schemas, tripped caps) throw errors that ALWAYS kill the script — they never dissolve into a per-item `null`. Constraints: concurrency and total-agent caps apply; no filesystem, network, timers, or Node.js APIs are provided — the agents do the work, the script only coordinates them. The run executes in the foreground: this call returns when the whole script finishes. */\n workflow: {\n /** The plain-JS workflow script body (top-level await allowed; NO `export const meta` statement; end with `return `). */\n script: string;\n /** The workflow identity block (plain JSON — never code). */\n meta: {\n /** Short kebab-case workflow name. */\n name: string;\n /** One-line description of what the workflow does. */\n description: string;\n /** Optional guidance on when this workflow applies. */\n whenToUse?: string;\n /** Optional phase declarations matched by phase() calls. */\n phases?: ({\n /** The phase title phase() calls match by exact string. */\n title: string;\n /** Optional one-line description of the phase. */\n detail?: string;\n /** Optional provider override this phase is expected to use. */\n provider?: string;\n /** Optional model override this phase is expected to use. */\n model?: string;\n } & Record)[];\n } & Record;\n /** Optional JSON input exposed to the script as the `args` global (wrap a bare list as a field, e.g. {\"files\": [...]}). */\n args?: Record;\n } & Record;\n /** Create or fully replace a UTF-8 text file. */\n write: {\n /** Path to write, resolved by the filesystem backend. */\n file_path: string;\n /** Full UTF-8 text content to write. */\n content: string;\n } & Record;\n}\n\ninterface ToolOutputMap {\n bash: {\n kind: \"background\";\n taskId: string;\n } | {\n kind: \"foreground\";\n exitCode: number | null;\n signal: string | null;\n timedOut: boolean;\n aborted: boolean;\n timeoutMs: number;\n stdout: {\n text: string;\n truncated: boolean;\n spillPath?: string;\n };\n stderr: {\n text: string;\n truncated: boolean;\n spillPath?: string;\n };\n sandbox?: {\n mode: string;\n denied: boolean;\n enforcement?: string;\n runnerFailed?: boolean;\n };\n };\n cordis_inspect: string;\n cordis_mount: {\n id: string;\n pluginName: string;\n state: \"pending\" | \"loading\" | \"active\" | \"failed\" | \"disposed\" | \"unloading\";\n provides: string[];\n waitingFor: string[];\n };\n cordis_unmount: {\n id: string;\n pluginName: string;\n };\n edit: {\n path: string;\n before: string;\n after: string;\n };\n ralph: {\n runId: string;\n agentsStarted: number;\n result: JsonValue;\n };\n read: {\n path: string;\n offset: number;\n lines: {\n number: number;\n text: string;\n }[];\n totalLines: number;\n };\n skill: {\n name: string;\n provider: string;\n resourceBase?: {\n kind: \"directory\";\n path: string;\n } | {\n kind: \"url\";\n url: string;\n } | {\n kind: \"opaque\";\n description: string;\n };\n content: string;\n };\n subagent: {\n kind: \"background\";\n taskId: string;\n } | {\n kind: \"foreground\";\n runId: string;\n output: JsonValue[];\n };\n subagent_fork: {\n kind: \"background\";\n taskId: string;\n } | {\n kind: \"foreground\";\n runId: string;\n output: JsonValue[];\n };\n task_kill: {\n outcome: \"cancellation-requested\" | \"already-finished\";\n task: {\n id: string;\n kind: string;\n label: string;\n status: \"running\" | \"stopping\" | \"completed\" | \"killed\" | \"failed\";\n detail?: string;\n startedAt: number;\n finishedAt?: number;\n };\n };\n task_list: ({\n id: string;\n kind: string;\n label: string;\n status: \"running\" | \"stopping\" | \"completed\" | \"killed\" | \"failed\";\n detail?: string;\n startedAt: number;\n finishedAt?: number;\n })[];\n task_output: {\n text: string;\n task: {\n id: string;\n kind: string;\n label: string;\n status: \"running\" | \"stopping\" | \"completed\" | \"killed\" | \"failed\";\n detail?: string;\n startedAt: number;\n finishedAt?: number;\n };\n };\n todo_write: {\n todos: ({\n content: string;\n status: \"pending\" | \"in_progress\" | \"completed\";\n })[];\n counts: {\n pending: number;\n inProgress: number;\n completed: number;\n };\n };\n workflow: {\n runId: string;\n agentsStarted: number;\n result: JsonValue;\n };\n write: {\n path: string;\n operation: \"create\" | \"update\";\n before: string | null;\n after: string;\n };\n}\n\ntype ToolName = keyof ToolOutputMap\n\ndeclare class ToolCallError extends Error {\n readonly name: \"ToolCallError\";\n readonly toolName: ToolName;\n}\n\ndeclare const tools: {\n [K in ToolName]: (args: ToolArgsMap[K]) => Promise;\n}\n```","tools":[{"name":"bash","description":"Execute a bash command (`bash -c`) and return its stdout/stderr. Each call runs in a fresh shell: no state (cwd, variables, functions) persists between calls — pass `workdir` instead of using `cd`. Non-zero exits are reported as `[exit code: N]`. Current harness environment facts are exposed through managed `$DSH_*` variables; inspect them when needed. Commands may run under a file sandbox; a blocked file operation is reported as `[sandbox: file access denied under mode]` — a policy denial, not a bug in the command; do not retry another way. Long output is truncated to its tail; the full output is saved to a file whose path is reported when available. Set `run_in_background: true` for long-running commands: the call returns a task id immediately; read its output with `task_output` and stop it with `task_kill`.","parameters":{"type":"object","properties":{"command":{"type":"string","description":"The bash command to execute."},"description":{"type":"string","description":"Clear, concise description of what this command does in active voice, 5-10 words (shown in the UI). Examples: \"ls\" → \"List files in current directory\"; \"git status\" → \"Show working tree status\"; \"npm install\" → \"Install package dependencies\"."},"timeoutMs":{"type":"number","description":"Timeout in milliseconds. The executor applies its configured default and cap, and kills the command on expiry."},"workdir":{"type":"string","description":"Working directory for this command. Defaults to the session workspace; a relative path is resolved against it."},"run_in_background":{"type":"boolean","description":"Run in the background and return a task id immediately (collect with task_output, stop with task_kill). No timeout applies."}},"required":["command","description"]}},{"name":"cordis_inspect","description":"Inspect the live Cordis runtime in the current DSH process. Read-only. Sections: `services` (every provided ctx service and the plugin fiber that owns it), `plugins` (all live plugin fibers with their lifecycle states), `tools` (the model-facing tools currently registered, i.e. what you can call), `temporary` (only temporary Plugins created by cordis_mount: id, name, state, provided services, awaited services, and lifetime), `api` (method signatures AND argument/return type shapes for every LIVE service — read this before writing plugin code that calls a service), `events` (every harness event with its dispatch mode and exact signature — pick listener targets here). Temporary Plugins exist only in memory, remain active across later turns, and disappear after cordis_unmount, toolset unload, or DSH restart; they are not restored automatically. The `temporary` section is a subset of `plugins`. Omit `what` to get all six sections. With `what:\"api\"` or `what:\"events\"`, pass an exact `name` to narrow to one service/event and include its original source JSDoc.","parameters":{"type":"object","properties":{"what":{"type":"string","description":"Limit the report to one section. Omit for all sections.","enum":["services","plugins","tools","temporary","api","events"]},"name":{"type":"string","description":"Exact service key or event name whose original JSDoc to include; valid only with what:\"api\" or what:\"events\"."}}}},{"name":"cordis_mount","description":"Mount a temporary Cordis Plugin in the current DSH process. This creates an in-memory runtime Plugin, not an installed or configured Plugin. It remains active across later turns until cordis_unmount, toolset unload, or DSH restart. It does not create files, install a package, change cordis.yml or personal/project config, survive restart, or automatically become permanent. To keep it, ask the Agent to implement a normal local, project, or repository Plugin through the regular development workflow. It may affect other sessions in the same process; the sandbox is not a security boundary, and injected services reach the real runtime. `code` runs now as the body of an async JavaScript function in an isolated sandbox and MUST `return` a plugin. Two forms: FUNCTION form `return (ctx) => { … }` — declares no inject, so it can register tools, listen to events, and provide services, but reaching ANY service (e.g. ctx.bash) throws; use it only when you need no services. OBJECT form `return { name?, inject: ['bash', 'llm', …], apply(ctx) { … } }` — declares dependencies, and cordis activates the plugin only after the services exist; PREFER this form. You may reach ONLY the services you list in inject: an undeclared service throws even if it exists, because an undeclared dependency would not be cleaned up if its provider is unmounted. BEFORE calling a service from your code, read cordis_inspect what:\"api\" — it lists method signatures AND the type shapes of their arguments/returns (do not guess a field's type; e.g. a bash run's stdout is an object, not a string). Inside `apply`, use the standard cordis API: `ctx.on(event, listener)` to observe events (see cordis_inspect what:\"events\"), or call `harness.registerTool(ctx, harness.defineTool({ name, description, parameters: { text: { type: 'string', required: true } }, output: { schema: { type: 'string' }, render(_args, value) { return [{ type: 'text', text: value }] } }, async execute(args) { return args.text } }))` to give yourself a new tool — it becomes callable on your NEXT step. Tool parameters: each key IS a property — { type: 'string'|'number'|'integer'|'boolean'|'null'|'object'|'array'|'json', required?: true, description?, enum?, const?, items?, properties? }; every direct DSL object declares additionalProperties: true|false, and oneOf: [schema, schema, ...] replaces type for an exact-one union. A raw JSON-Schema { type: 'object', properties, required?: […] } wrapper is also accepted with open-by-default objects. A tool's `execute` MUST return the lossless JSON value declared by `output.schema`; `output.render(args, value)` separately returns Native/model content blocks. Temporary Plugins can COMPOSE: one Plugin may `ctx.provide('name', value)` a service and another may declare `inject: ['name']` to consume it — the consumer stays pending until the provider exists and returns to pending when the provider is unmounted. Everything registered inside `apply` is cleaned up automatically by cordis_unmount. Sandbox globals: `console` (tagged `[cordis:]`, writes through to the harness terminal), `harness.defineTool`, `harness.registerTool`, `btoa`, `atob`, `TextEncoder`, `TextDecoder`. Node APIs are DISABLED — do filesystem/network/timer work through the cordis services, never Node built-ins: `require`, `setTimeout`/`setInterval`, and `fetch` throw redirect errors; `process` and `Buffer` are undefined. Instead use inject: ['fs'] + ctx.fs for files, inject: ['web'] + ctx.web for HTTP, inject: ['bash'] + ctx.bash for processes, and inject: ['timer'] + ctx.setTimeout/ctx.setInterval for timing (fiber effects, auto-cleaned when unmounted) — cordis_inspect what:\"api\" shows what THIS runtime provides. Write PLAIN JavaScript, not TypeScript (no `as`, no type annotations). Cautions: (1) waterfall events (e.g. tools/pre-execute) hand the listener a trailing `next` callback which MUST be called — returning without `next()` VETOES the call; prefer plain notification events unless you intend to intercept. (2) Never await something that only resolves after the current turn (your code runs INSIDE a tool call of that turn — it would deadlock). (3) Your `ctx` is a restricted façade: you can register tools, observe events, provide/consume services, and use timers, but framework internals (ctx.root, ctx.fiber, ctx.extend, ctx.plugin, …) are withheld. It is not a security boundary though — the services you inject (e.g. ctx.bash) reach the real runtime.","parameters":{"type":"object","properties":{"code":{"type":"string","description":"JavaScript body returning a temporary Plugin; evaluated now and saved nowhere."}},"required":["code"]}},{"name":"cordis_unmount","description":"Unmount a current-process temporary Plugin created by cordis_mount. Waits for its tools, listeners, services, timers, and other owned effects to clean up completely. Only dyn-N temporary ids are accepted; this cannot remove Loader, configured, or installed Plugins.","parameters":{"type":"object","properties":{"id":{"type":"string","description":"The temporary Plugin id returned by cordis_mount (for example \"dyn-1\"); valid only in this process and invalid after unmount or restart."}},"required":["id"]}},{"name":"edit","description":"Edit an existing UTF-8 text file by replacing literal text.","parameters":{"type":"object","properties":{"file_path":{"type":"string","description":"Path to edit, resolved by the filesystem backend."},"old_string":{"type":"string","description":"Literal text to replace. Must match exactly."},"new_string":{"type":"string","description":"Literal replacement text. Use an empty string to delete the match."},"replace_all":{"type":"boolean","description":"Replace all matches. Defaults to false; when false, old_string must appear exactly once."}},"required":["file_path","old_string","new_string"]}},{"name":"ralph","description":"Run a foreground fresh-agent Ralph loop toward one immutable objective. Use only when the direct human explicitly asks for Ralph or fresh-agent iteration. Each round opens a new child with no parent conversation or prior child session; the shared workspace is long-term memory, and only a bounded structured report crosses rounds. The call returns when a worker reports completion or a concrete blocker, or at the round limit. Ordinary long-running same-session work belongs to goal tools.","parameters":{"type":"object","properties":{"objective":{"type":"string","description":"The immutable completion objective for every fresh Ralph round."},"maxRounds":{"type":"number","description":"Optional positive safe-integer round cap, bounded by the deployment ceiling."}},"required":["objective"]}},{"name":"read","description":"Read a UTF-8 text file and return line-numbered content.","parameters":{"type":"object","properties":{"file_path":{"type":"string","description":"Path to read, resolved by the filesystem backend."},"offset":{"type":"number","description":"1-based first line to return. Defaults to 1."},"limit":{"type":"number","description":"Maximum number of lines to return. Defaults to 2000."}},"required":["file_path"]}},{"name":"run_code","description":"Execute a TypeScript program against the available tools. Write the BODY of an async function (erasable syntax only; top-level `await` and `return` work) and call tools as `await tools.name(args)` per the declarations in the system prompt. Only what you print or return comes back — curate it.","parameters":{"type":"object","properties":{"code":{"type":"string","description":"The program: the body of an async TypeScript function."},"description":{"type":"string","description":"Clear, concise description of what this program does in active voice, 5-10 words (shown in the UI). Examples: \"Count TODO markers across packages\"; \"Read failing test and its fixture\"; \"Rename config key in every cordis.yml\"."}},"required":["code","description"]}},{"name":"skill","description":"Load the full instructions for an available skill. Call this with the exact skill name from the session skill catalog before acting on a task that names or clearly matches that skill.","parameters":{"type":"object","properties":{"name":{"type":"string","description":"The exact skill name from the available skills list."}},"required":["name"]}},{"name":"subagent","description":"Delegate a self-contained task to a subagent (a separate agent that works in its own context) and return its final result. Use this to offload focused, independent work — research, a scoped implementation, an analysis — so it does not consume this conversation's context. The subagent runs to completion and you receive only its final answer, not its intermediate steps. Give it a complete, standalone prompt: it does not see this conversation. Set `run_in_background: true` to return a task id; collect with `task_output` and stop with `task_kill`.","parameters":{"type":"object","properties":{"description":{"type":"string","description":"A short (3-5 word) description of the delegated task, for display."},"prompt":{"type":"string","description":"The complete, self-contained task for the subagent. It does not share this conversation's context, so include everything it needs."},"run_in_background":{"type":"boolean","description":"Run as a background task and return its id; collect with task_output or stop with task_kill."}},"required":["description","prompt"]}},{"name":"subagent_fork","description":"Delegate a task to a subagent that inherits this conversation: a child agent seeded with all completed turns so far (it does not see the current in-flight turn), returning only its final result. Use this when the subtask builds on this conversation's context — a follow-up analysis, a review, a continuation — without consuming this conversation's context for the work itself. You receive only its final answer, not its intermediate steps. Set `run_in_background: true` to return a task id; collect with `task_output` and stop with `task_kill`.","parameters":{"type":"object","properties":{"description":{"type":"string","description":"A short (3-5 word) description of the delegated task, for display."},"prompt":{"type":"string","description":"The task for the subagent. It already sees this conversation's completed turns, so build on them freely and state only what is new."},"run_in_background":{"type":"boolean","description":"Run as a background task and return its id; collect with task_output or stop with task_kill."}},"required":["description","prompt"]}},{"name":"task_kill","description":"Request cancellation of a running background task by task id. Returns immediately; the task settles as killed once its work actually stops.","parameters":{"type":"object","properties":{"task_id":{"type":"string","description":"Task id returned by the tool that started the background work."},"reason":{"type":"string","description":"Optional short reason, recorded in the log and forwarded to the task."}},"required":["task_id"]}},{"name":"task_list","description":"List your background tasks (running and finished) with their ids, kinds, and statuses.","parameters":{"type":"object","properties":{}}},{"name":"task_output","description":"Read a background task. Stream tasks return only output since the previous read; final-output tasks return their result after settlement. Every response ends with `[status: ...]`. Reads are non-blocking unless `wait: true`, which waits up to the configured cap.","parameters":{"type":"object","properties":{"task_id":{"type":"string","description":"Task id returned by the tool that started the background work."},"wait":{"type":"boolean","description":"Block until the task reaches a terminal status or the timeout expires. A timed-out wait returns [status: running] and leaves the task alive."},"timeout_ms":{"type":"number","description":"Max wait in milliseconds (only meaningful with wait: true). Defaults to the configured wait timeout; capped by the configured maximum."}},"required":["task_id"]}},{"name":"todo_write","description":"Record and update a structured task list for the current work. Send the ENTIRE list every call — it REPLACES the previous list (there are no partial updates, no per-item edits). Use it to plan multi-step work and show progress: add one todo per concrete step before you start. Keep AT MOST ONE todo `in_progress` at a time; while work remains, exactly one active task should be `in_progress`. Mark a todo `completed` the moment it is done (do not batch completions), and allow no `in_progress` item only once all work is complete. Skip the list for trivial single-step tasks. Statuses: `pending` (not started), `in_progress` (being worked on now), `completed` (finished).","parameters":{"type":"object","properties":{"todos":{"type":"array","description":"The COMPLETE task list, replacing any previous list.","items":{"type":"object","additionalProperties":false,"properties":{"content":{"type":"string","description":"What the task is — a short imperative line."},"status":{"type":"string","description":"pending (not started) | in_progress (now) | completed (done).","enum":["pending","in_progress","completed"]}},"required":["content","status"]}}},"required":["todos"]}},{"name":"workflow","description":"Run a JavaScript workflow script that orchestrates subagents at scale. Use this for work that fans out across many independent pieces — an audit over many files, a migration, multi-angle research, adversarial verification of findings — where you write the orchestration as a script instead of delegating turn by turn.\n\nThe workflow's identity rides the `meta` parameter as JSON: required `name` (short kebab-case) and `description` strings, optional `whenToUse` string and `phases` array (`{title, detail?, provider?, model?}`). The `script` parameter is the plain JavaScript body ONLY (NOT TypeScript, and NO `export const meta` statement — meta is a parameter, not code), running with top-level await; end with `return ` — the value must be JSON-serializable and is this tool's result.\n\nScript-body hooks:\n- `agent(prompt, opts?): Promise` — run one subagent to completion. Without `opts.schema` it resolves to the child's final text; with `opts.schema` (an object-rooted JSON Schema using ONLY type/properties/required/additionalProperties/items/enum/const/oneOf — no pattern/format/numeric bounds) it resolves to the validated object. Resolves `null` when the child fails (filter with `.filter(Boolean)`). Other opts: `label` (display), `phase` (progress group), and independent `provider`/`model` LLM target overrides (either may be provided alone). Anything else (`effort`/`isolation`/`agentType`) is rejected loudly.\n- `pipeline(items, ...stages): Promise` — run each item through the stages independently with NO barrier between stages (prefer this for multi-stage work). Each stage receives `(prev, item, index)`. An ordinary stage throw drops that ITEM to `null` and skips its remaining stages.\n- `parallel(thunks): Promise` — run zero-argument functions concurrently and await ALL of them (a barrier; use only when a stage genuinely needs every prior result together). A throwing thunk resolves to `null`.\n- `phase(title)` — start a progress phase; `log(message)` — narrate progress; `args` — the tool call's `args` input, verbatim.\n\nMisused hooks (bad arguments, unknown options, unsupported schemas, tripped caps) throw errors that ALWAYS kill the script — they never dissolve into a per-item `null`.\n\nConstraints: concurrency and total-agent caps apply; no filesystem, network, timers, or Node.js APIs are provided — the agents do the work, the script only coordinates them. The run executes in the foreground: this call returns when the whole script finishes.","parameters":{"type":"object","properties":{"script":{"type":"string","description":"The plain-JS workflow script body (top-level await allowed; NO `export const meta` statement; end with `return `)."},"meta":{"type":"object","description":"The workflow identity block (plain JSON — never code).","additionalProperties":true,"properties":{"name":{"type":"string","description":"Short kebab-case workflow name."},"description":{"type":"string","description":"One-line description of what the workflow does."},"whenToUse":{"type":"string","description":"Optional guidance on when this workflow applies."},"phases":{"type":"array","description":"Optional phase declarations matched by phase() calls.","items":{"type":"object","additionalProperties":true,"properties":{"title":{"type":"string","description":"The phase title phase() calls match by exact string."},"detail":{"type":"string","description":"Optional one-line description of the phase."},"provider":{"type":"string","description":"Optional provider override this phase is expected to use."},"model":{"type":"string","description":"Optional model override this phase is expected to use."}},"required":["title"]}}},"required":["name","description"]},"args":{"type":"object","description":"Optional JSON input exposed to the script as the `args` global (wrap a bare list as a field, e.g. {\"files\": [...]}).","additionalProperties":true}},"required":["script","meta"]}},{"name":"write","description":"Create or fully replace a UTF-8 text file.","parameters":{"type":"object","properties":{"file_path":{"type":"string","description":"Path to write, resolved by the filesystem backend."},"content":{"type":"string","description":"Full UTF-8 text content to write."}},"required":["file_path","content"]}}]},"reason":"initial"}} {"type":"assistant/chunk","seq":5,"time":1783950001005,"data":{"turn":1,"step":1,"chunk":{"type":"block-start","index":0,"blockType":"text"}}} {"type":"assistant/chunk","seq":6,"time":1783957884564,"data":{"turn":1,"step":1,"chunk":{"type":"text-delta","index":0,"text":"DIRECT_CHILD_OK"}}} {"type":"assistant/chunk","seq":7,"time":1783957884564,"data":{"turn":1,"step":1,"chunk":{"type":"block-end","index":0,"block":{"type":"text","text":"DIRECT_CHILD_OK"}}}} diff --git a/examples/headless-agent/tests/snapshots/advanced-toolchain/session.2.jsonl b/examples/headless-agent/tests/snapshots/advanced-toolchain/session.2.jsonl index 7a12caea56..370ee495cb 100644 --- a/examples/headless-agent/tests/snapshots/advanced-toolchain/session.2.jsonl +++ b/examples/headless-agent/tests/snapshots/advanced-toolchain/session.2.jsonl @@ -1,9 +1,9 @@ -{"type":"session","version":0,"id":"33333333-3333-4333-8333-333333333333","createdAt":1783950002000,"cwd":"/tmp/advanced-headless","parentSession":"11111111-1111-4111-8111-111111111111","delegationDepth":1} +{"type":"session","version":0,"id":"33333333-3333-4333-8333-333333333333","createdAt":1783950002000,"cwd":"{{cwd}}","parentSession":"11111111-1111-4111-8111-111111111111","delegationDepth":1} {"type":"turn/start","seq":0,"time":1783957884700,"data":{"turn":1,"trigger":{"kind":"message","source":{"kind":"user"}}}} {"type":"user/message","seq":1,"time":1783957884700,"data":{"content":[{"type":"text","text":"Reply with exactly WORKFLOW_CHILD_OK and nothing else."}],"source":{"kind":"user"},"role":"user","id":"1d565b63-5689-4c09-9686-abd3ee379e28"},"surfaceOp":"append"} {"type":"session/title","seq":2,"time":1783957884700,"data":{"title":"Reply with exactly WORKFLOW_CHILD_OK and","messageSeqs":[1],"source":{"kind":"fallback"}}} {"type":"step/start","seq":3,"time":1783957884700,"data":{"turn":1,"step":1}} -{"type":"request/header","seq":4,"time":1783957884701,"data":{"header":{"config":{"provider":"deepseek","model":"deepseek-v4-flash"},"system":"You are an AI agent powered by the DeepSeek Harness SDK.\n\nYou are headless-agent, a coding assistant powered by the deepseek-v4-flash model. Your working directory is /tmp/advanced-headless.\n\nVerify your work by running the code or tests. Keep answers brief and factual.\n\n\nUse the read tool — not shell commands like cat — to inspect text files. Results include line numbers. Use offset and limit to continue reading large files.\n\nUse the write tool to create files or completely replace file contents. Existing files are overwritten, so read an existing file first (the default fs-policy requires it) and prefer edit for targeted changes.\n\nUse the edit tool for targeted changes to existing UTF-8 text files. It replaces literal old_string with new_string; by default old_string must appear exactly once. If old_string appears multiple times, provide a more specific old_string or set replace_all to true. Read the file first (the default fs-policy requires it), unless you just created or edited it in this session.\n\nCheck the [exit code: N] marker on every bash result; investigate failures before moving on.\n\nTrack every background task id you start. You are notified in-session when a task finishes — do not busy-poll or sleep on one; keep working on independent steps and do not duplicate a running task's work. Before giving a final answer, collect every still-relevant task with task_output (set wait: true only when you are genuinely blocked on it), and task_kill tasks that stopped mattering.\n\nUse the workflow tool ONLY when the user explicitly asks for a workflow or for large multi-agent orchestration: you write a JavaScript script (the tool description documents the exact format) that fans work out across many subagents with phases and structured results. For one or two delegations, prefer plain subagent calls.\n\nUse the ralph tool ONLY when the direct human explicitly asks for a Ralph loop or fresh-agent iterative execution. Each Ralph round starts a fresh child with no conversation seed and uses the shared workspace as durable memory. Completion and blockers are worker reports, not independent evaluation. Use same-session goal tools for ordinary long-running objectives, and plain subagents or workflows for bounded delegation and fan-out.\n\n## Writing code for run_code\n\nPass `run_code` the body of an async TypeScript function (erasable syntax only — no `enum` or namespaces; type annotations are advisory, the code runs type-stripped). Inside the program:\n\n- Call tools as `await tools.name(args)` — quoted access for exotic names: `tools[\"my-tool\"](args)`. Every call resolves to the tool's typed canonical JSON value. Tool arguments must be lossless JSON.\n- A FAILED tool call rejects with `ToolCallError`, whose `toolName` identifies the failed tool and whose `message` is human-readable — `try/catch` it to handle and continue.\n- Independent read-only calls MAY overlap under `Promise.all` (safe calls run concurrently; mutating calls run alone, in submission order). Sequence dependent work with `await`.\n- Emit results with `return` and/or `console.log(...)`. ONLY what you print or return comes back to you — intermediate tool results never enter the conversation, so extract just what you need.\n\nThe available tools:\n\n```ts\ntype JsonValue = null | boolean | number | string | JsonValue[] | { [key: string]: JsonValue }\n\ninterface ToolArgsMap {\n /** Execute a bash command (`bash -c`) and return its stdout/stderr. Each call runs in a fresh shell: no state (cwd, variables, functions) persists between calls — pass `workdir` instead of using `cd`. Non-zero exits are reported as `[exit code: N]`. Current harness environment facts are exposed through managed `$DSH_*` variables; inspect them when needed. Commands may run under a file sandbox; a blocked file operation is reported as `[sandbox: file access denied under mode]` — a policy denial, not a bug in the command; do not retry another way. Long output is truncated to its tail; the full output is saved to a file whose path is reported when available. Set `run_in_background: true` for long-running commands: the call returns a task id immediately; read its output with `task_output` and stop it with `task_kill`. */\n bash: {\n /** The bash command to execute. */\n command: string;\n /** Clear, concise description of what this command does in active voice, 5-10 words (shown in the UI). Examples: \"ls\" → \"List files in current directory\"; \"git status\" → \"Show working tree status\"; \"npm install\" → \"Install package dependencies\". */\n description: string;\n /** Timeout in milliseconds. The executor applies its configured default and cap, and kills the command on expiry. */\n timeoutMs?: number;\n /** Working directory for this command. Defaults to the session workspace; a relative path is resolved against it. */\n workdir?: string;\n /** Run in the background and return a task id immediately (collect with task_output, stop with task_kill). No timeout applies. */\n run_in_background?: boolean;\n } & Record;\n /** Inspect the live Cordis runtime in the current DSH process. Read-only. Sections: `services` (every provided ctx service and the plugin fiber that owns it), `plugins` (all live plugin fibers with their lifecycle states), `tools` (the model-facing tools currently registered, i.e. what you can call), `temporary` (only temporary Plugins created by cordis_mount: id, name, state, provided services, awaited services, and lifetime), `api` (method signatures AND argument/return type shapes for every LIVE service — read this before writing plugin code that calls a service), `events` (every harness event with its dispatch mode and exact signature — pick listener targets here). Temporary Plugins exist only in memory, remain active across later turns, and disappear after cordis_unmount, toolset unload, or DSH restart; they are not restored automatically. The `temporary` section is a subset of `plugins`. Omit `what` to get all six sections. With `what:\"api\"` or `what:\"events\"`, pass an exact `name` to narrow to one service/event and include its original source JSDoc. */\n cordis_inspect: {\n /** Limit the report to one section. Omit for all sections. */\n what?: \"services\" | \"plugins\" | \"tools\" | \"temporary\" | \"api\" | \"events\";\n /** Exact service key or event name whose original JSDoc to include; valid only with what:\"api\" or what:\"events\". */\n name?: string;\n } & Record;\n /** Mount a temporary Cordis Plugin in the current DSH process. This creates an in-memory runtime Plugin, not an installed or configured Plugin. It remains active across later turns until cordis_unmount, toolset unload, or DSH restart. It does not create files, install a package, change cordis.yml or personal/project config, survive restart, or automatically become permanent. To keep it, ask the Agent to implement a normal local, project, or repository Plugin through the regular development workflow. It may affect other sessions in the same process; the sandbox is not a security boundary, and injected services reach the real runtime. `code` runs now as the body of an async JavaScript function in an isolated sandbox and MUST `return` a plugin. Two forms: FUNCTION form `return (ctx) => { … }` — declares no inject, so it can register tools, listen to events, and provide services, but reaching ANY service (e.g. ctx.bash) throws; use it only when you need no services. OBJECT form `return { name?, inject: ['bash', 'llm', …], apply(ctx) { … } }` — declares dependencies, and cordis activates the plugin only after the services exist; PREFER this form. You may reach ONLY the services you list in inject: an undeclared service throws even if it exists, because an undeclared dependency would not be cleaned up if its provider is unmounted. BEFORE calling a service from your code, read cordis_inspect what:\"api\" — it lists method signatures AND the type shapes of their arguments/returns (do not guess a field's type; e.g. a bash run's stdout is an object, not a string). Inside `apply`, use the standard cordis API: `ctx.on(event, listener)` to observe events (see cordis_inspect what:\"events\"), or call `harness.registerTool(ctx, harness.defineTool({ name, description, parameters: { text: { type: 'string', required: true } }, output: { schema: { type: 'string' }, render(_args, value) { return [{ type: 'text', text: value }] } }, async execute(args) { return args.text } }))` to give yourself a new tool — it becomes callable on your NEXT step. Tool parameters: each key IS a property — { type: 'string'|'number'|'integer'|'boolean'|'null'|'object'|'array'|'json', required?: true, description?, enum?, const?, items?, properties? }; every direct DSL object declares additionalProperties: true|false, and oneOf: [schema, schema, ...] replaces type for an exact-one union. A raw JSON-Schema { type: 'object', properties, required?: […] } wrapper is also accepted with open-by-default objects. A tool's `execute` MUST return the lossless JSON value declared by `output.schema`; `output.render(args, value)` separately returns Native/model content blocks. Temporary Plugins can COMPOSE: one Plugin may `ctx.provide('name', value)` a service and another may declare `inject: ['name']` to consume it — the consumer stays pending until the provider exists and returns to pending when the provider is unmounted. Everything registered inside `apply` is cleaned up automatically by cordis_unmount. Sandbox globals: `console` (tagged `[cordis:]`, writes through to the harness terminal), `harness.defineTool`, `harness.registerTool`, `btoa`, `atob`, `TextEncoder`, `TextDecoder`. Node APIs are DISABLED — do filesystem/network/timer work through the cordis services, never Node built-ins: `require`, `setTimeout`/`setInterval`, and `fetch` throw redirect errors; `process` and `Buffer` are undefined. Instead use inject: ['fs'] + ctx.fs for files, inject: ['web'] + ctx.web for HTTP, inject: ['bash'] + ctx.bash for processes, and inject: ['timer'] + ctx.setTimeout/ctx.setInterval for timing (fiber effects, auto-cleaned when unmounted) — cordis_inspect what:\"api\" shows what THIS runtime provides. Write PLAIN JavaScript, not TypeScript (no `as`, no type annotations). Cautions: (1) waterfall events (e.g. tools/pre-execute) hand the listener a trailing `next` callback which MUST be called — returning without `next()` VETOES the call; prefer plain notification events unless you intend to intercept. (2) Never await something that only resolves after the current turn (your code runs INSIDE a tool call of that turn — it would deadlock). (3) Your `ctx` is a restricted façade: you can register tools, observe events, provide/consume services, and use timers, but framework internals (ctx.root, ctx.fiber, ctx.extend, ctx.plugin, …) are withheld. It is not a security boundary though — the services you inject (e.g. ctx.bash) reach the real runtime. */\n cordis_mount: {\n /** JavaScript body returning a temporary Plugin; evaluated now and saved nowhere. */\n code: string;\n } & Record;\n /** Unmount a current-process temporary Plugin created by cordis_mount. Waits for its tools, listeners, services, timers, and other owned effects to clean up completely. Only dyn-N temporary ids are accepted; this cannot remove Loader, configured, or installed Plugins. */\n cordis_unmount: {\n /** The temporary Plugin id returned by cordis_mount (for example \"dyn-1\"); valid only in this process and invalid after unmount or restart. */\n id: string;\n } & Record;\n /** Edit an existing UTF-8 text file by replacing literal text. */\n edit: {\n /** Path to edit, resolved by the filesystem backend. */\n file_path: string;\n /** Literal text to replace. Must match exactly. */\n old_string: string;\n /** Literal replacement text. Use an empty string to delete the match. */\n new_string: string;\n /** Replace all matches. Defaults to false; when false, old_string must appear exactly once. */\n replace_all?: boolean;\n } & Record;\n /** Run a foreground fresh-agent Ralph loop toward one immutable objective. Use only when the direct human explicitly asks for Ralph or fresh-agent iteration. Each round opens a new child with no parent conversation or prior child session; the shared workspace is long-term memory, and only a bounded structured report crosses rounds. The call returns when a worker reports completion or a concrete blocker, or at the round limit. Ordinary long-running same-session work belongs to goal tools. */\n ralph: {\n /** The immutable completion objective for every fresh Ralph round. */\n objective: string;\n /** Optional positive safe-integer round cap, bounded by the deployment ceiling. */\n maxRounds?: number;\n } & Record;\n /** Read a UTF-8 text file and return line-numbered content. */\n read: {\n /** Path to read, resolved by the filesystem backend. */\n file_path: string;\n /** 1-based first line to return. Defaults to 1. */\n offset?: number;\n /** Maximum number of lines to return. Defaults to 2000. */\n limit?: number;\n } & Record;\n /** Load the full instructions for an available skill. Call this with the exact skill name from the session skill catalog before acting on a task that names or clearly matches that skill. */\n skill: {\n /** The exact skill name from the available skills list. */\n name: string;\n } & Record;\n /** Delegate a self-contained task to a subagent (a separate agent that works in its own context) and return its final result. Use this to offload focused, independent work — research, a scoped implementation, an analysis — so it does not consume this conversation's context. The subagent runs to completion and you receive only its final answer, not its intermediate steps. Give it a complete, standalone prompt: it does not see this conversation. Set `run_in_background: true` to return a task id; collect with `task_output` and stop with `task_kill`. */\n subagent: {\n /** A short (3-5 word) description of the delegated task, for display. */\n description: string;\n /** The complete, self-contained task for the subagent. It does not share this conversation's context, so include everything it needs. */\n prompt: string;\n /** Run as a background task and return its id; collect with task_output or stop with task_kill. */\n run_in_background?: boolean;\n } & Record;\n /** Delegate a task to a subagent that inherits this conversation: a child agent seeded with all completed turns so far (it does not see the current in-flight turn), returning only its final result. Use this when the subtask builds on this conversation's context — a follow-up analysis, a review, a continuation — without consuming this conversation's context for the work itself. You receive only its final answer, not its intermediate steps. Set `run_in_background: true` to return a task id; collect with `task_output` and stop with `task_kill`. */\n subagent_fork: {\n /** A short (3-5 word) description of the delegated task, for display. */\n description: string;\n /** The task for the subagent. It already sees this conversation's completed turns, so build on them freely and state only what is new. */\n prompt: string;\n /** Run as a background task and return its id; collect with task_output or stop with task_kill. */\n run_in_background?: boolean;\n } & Record;\n /** Request cancellation of a running background task by task id. Returns immediately; the task settles as killed once its work actually stops. */\n task_kill: {\n /** Task id returned by the tool that started the background work. */\n task_id: string;\n /** Optional short reason, recorded in the log and forwarded to the task. */\n reason?: string;\n } & Record;\n /** List your background tasks (running and finished) with their ids, kinds, and statuses. */\n task_list: Record;\n /** Read a background task. Stream tasks return only output since the previous read; final-output tasks return their result after settlement. Every response ends with `[status: ...]`. Reads are non-blocking unless `wait: true`, which waits up to the configured cap. */\n task_output: {\n /** Task id returned by the tool that started the background work. */\n task_id: string;\n /** Block until the task reaches a terminal status or the timeout expires. A timed-out wait returns [status: running] and leaves the task alive. */\n wait?: boolean;\n /** Max wait in milliseconds (only meaningful with wait: true). Defaults to the configured wait timeout; capped by the configured maximum. */\n timeout_ms?: number;\n } & Record;\n /** Record and update a structured task list for the current work. Send the ENTIRE list every call — it REPLACES the previous list (there are no partial updates, no per-item edits). Use it to plan multi-step work and show progress: add one todo per concrete step before you start. Keep AT MOST ONE todo `in_progress` at a time; while work remains, exactly one active task should be `in_progress`. Mark a todo `completed` the moment it is done (do not batch completions), and allow no `in_progress` item only once all work is complete. Skip the list for trivial single-step tasks. Statuses: `pending` (not started), `in_progress` (being worked on now), `completed` (finished). */\n todo_write: {\n /** The COMPLETE task list, replacing any previous list. */\n todos: ({\n /** What the task is — a short imperative line. */\n content: string;\n /** pending (not started) | in_progress (now) | completed (done). */\n status: \"pending\" | \"in_progress\" | \"completed\";\n })[];\n } & Record;\n /** Run a JavaScript workflow script that orchestrates subagents at scale. Use this for work that fans out across many independent pieces — an audit over many files, a migration, multi-angle research, adversarial verification of findings — where you write the orchestration as a script instead of delegating turn by turn. The workflow's identity rides the `meta` parameter as JSON: required `name` (short kebab-case) and `description` strings, optional `whenToUse` string and `phases` array (`{title, detail?, provider?, model?}`). The `script` parameter is the plain JavaScript body ONLY (NOT TypeScript, and NO `export const meta` statement — meta is a parameter, not code), running with top-level await; end with `return ` — the value must be JSON-serializable and is this tool's result. Script-body hooks: - `agent(prompt, opts?): Promise` — run one subagent to completion. Without `opts.schema` it resolves to the child's final text; with `opts.schema` (an object-rooted JSON Schema using ONLY type/properties/required/additionalProperties/items/enum/const/oneOf — no pattern/format/numeric bounds) it resolves to the validated object. Resolves `null` when the child fails (filter with `.filter(Boolean)`). Other opts: `label` (display), `phase` (progress group), and independent `provider`/`model` LLM target overrides (either may be provided alone). Anything else (`effort`/`isolation`/`agentType`) is rejected loudly. - `pipeline(items, ...stages): Promise` — run each item through the stages independently with NO barrier between stages (prefer this for multi-stage work). Each stage receives `(prev, item, index)`. An ordinary stage throw drops that ITEM to `null` and skips its remaining stages. - `parallel(thunks): Promise` — run zero-argument functions concurrently and await ALL of them (a barrier; use only when a stage genuinely needs every prior result together). A throwing thunk resolves to `null`. - `phase(title)` — start a progress phase; `log(message)` — narrate progress; `args` — the tool call's `args` input, verbatim. Misused hooks (bad arguments, unknown options, unsupported schemas, tripped caps) throw errors that ALWAYS kill the script — they never dissolve into a per-item `null`. Constraints: concurrency and total-agent caps apply; no filesystem, network, timers, or Node.js APIs are provided — the agents do the work, the script only coordinates them. The run executes in the foreground: this call returns when the whole script finishes. */\n workflow: {\n /** The plain-JS workflow script body (top-level await allowed; NO `export const meta` statement; end with `return `). */\n script: string;\n /** The workflow identity block (plain JSON — never code). */\n meta: {\n /** Short kebab-case workflow name. */\n name: string;\n /** One-line description of what the workflow does. */\n description: string;\n /** Optional guidance on when this workflow applies. */\n whenToUse?: string;\n /** Optional phase declarations matched by phase() calls. */\n phases?: ({\n /** The phase title phase() calls match by exact string. */\n title: string;\n /** Optional one-line description of the phase. */\n detail?: string;\n /** Optional provider override this phase is expected to use. */\n provider?: string;\n /** Optional model override this phase is expected to use. */\n model?: string;\n } & Record)[];\n } & Record;\n /** Optional JSON input exposed to the script as the `args` global (wrap a bare list as a field, e.g. {\"files\": [...]}). */\n args?: Record;\n } & Record;\n /** Create or fully replace a UTF-8 text file. */\n write: {\n /** Path to write, resolved by the filesystem backend. */\n file_path: string;\n /** Full UTF-8 text content to write. */\n content: string;\n } & Record;\n}\n\ninterface ToolOutputMap {\n bash: {\n kind: \"background\";\n taskId: string;\n } | {\n kind: \"foreground\";\n exitCode: number | null;\n signal: string | null;\n timedOut: boolean;\n aborted: boolean;\n timeoutMs: number;\n stdout: {\n text: string;\n truncated: boolean;\n spillPath?: string;\n };\n stderr: {\n text: string;\n truncated: boolean;\n spillPath?: string;\n };\n sandbox?: {\n mode: string;\n denied: boolean;\n enforcement?: string;\n runnerFailed?: boolean;\n };\n };\n cordis_inspect: string;\n cordis_mount: {\n id: string;\n pluginName: string;\n state: \"pending\" | \"loading\" | \"active\" | \"failed\" | \"disposed\" | \"unloading\";\n provides: string[];\n waitingFor: string[];\n };\n cordis_unmount: {\n id: string;\n pluginName: string;\n };\n edit: {\n path: string;\n before: string;\n after: string;\n };\n ralph: {\n runId: string;\n agentsStarted: number;\n result: JsonValue;\n };\n read: {\n path: string;\n offset: number;\n lines: {\n number: number;\n text: string;\n }[];\n totalLines: number;\n };\n skill: {\n name: string;\n provider: string;\n resourceBase?: {\n kind: \"directory\";\n path: string;\n } | {\n kind: \"url\";\n url: string;\n } | {\n kind: \"opaque\";\n description: string;\n };\n content: string;\n };\n subagent: {\n kind: \"background\";\n taskId: string;\n } | {\n kind: \"foreground\";\n runId: string;\n output: JsonValue[];\n };\n subagent_fork: {\n kind: \"background\";\n taskId: string;\n } | {\n kind: \"foreground\";\n runId: string;\n output: JsonValue[];\n };\n task_kill: {\n outcome: \"cancellation-requested\" | \"already-finished\";\n task: {\n id: string;\n kind: string;\n label: string;\n status: \"running\" | \"stopping\" | \"completed\" | \"killed\" | \"failed\";\n detail?: string;\n startedAt: number;\n finishedAt?: number;\n };\n };\n task_list: ({\n id: string;\n kind: string;\n label: string;\n status: \"running\" | \"stopping\" | \"completed\" | \"killed\" | \"failed\";\n detail?: string;\n startedAt: number;\n finishedAt?: number;\n })[];\n task_output: {\n text: string;\n task: {\n id: string;\n kind: string;\n label: string;\n status: \"running\" | \"stopping\" | \"completed\" | \"killed\" | \"failed\";\n detail?: string;\n startedAt: number;\n finishedAt?: number;\n };\n };\n todo_write: {\n todos: ({\n content: string;\n status: \"pending\" | \"in_progress\" | \"completed\";\n })[];\n counts: {\n pending: number;\n inProgress: number;\n completed: number;\n };\n };\n workflow: {\n runId: string;\n agentsStarted: number;\n result: JsonValue;\n };\n write: {\n path: string;\n operation: \"create\" | \"update\";\n before: string | null;\n after: string;\n };\n}\n\ntype ToolName = keyof ToolOutputMap\n\ndeclare class ToolCallError extends Error {\n readonly name: \"ToolCallError\";\n readonly toolName: ToolName;\n}\n\ndeclare const tools: {\n [K in ToolName]: (args: ToolArgsMap[K]) => Promise;\n}\n```","tools":[{"name":"bash","description":"Execute a bash command (`bash -c`) and return its stdout/stderr. Each call runs in a fresh shell: no state (cwd, variables, functions) persists between calls — pass `workdir` instead of using `cd`. Non-zero exits are reported as `[exit code: N]`. Current harness environment facts are exposed through managed `$DSH_*` variables; inspect them when needed. Commands may run under a file sandbox; a blocked file operation is reported as `[sandbox: file access denied under mode]` — a policy denial, not a bug in the command; do not retry another way. Long output is truncated to its tail; the full output is saved to a file whose path is reported when available. Set `run_in_background: true` for long-running commands: the call returns a task id immediately; read its output with `task_output` and stop it with `task_kill`.","parameters":{"type":"object","properties":{"command":{"type":"string","description":"The bash command to execute."},"description":{"type":"string","description":"Clear, concise description of what this command does in active voice, 5-10 words (shown in the UI). Examples: \"ls\" → \"List files in current directory\"; \"git status\" → \"Show working tree status\"; \"npm install\" → \"Install package dependencies\"."},"timeoutMs":{"type":"number","description":"Timeout in milliseconds. The executor applies its configured default and cap, and kills the command on expiry."},"workdir":{"type":"string","description":"Working directory for this command. Defaults to the session workspace; a relative path is resolved against it."},"run_in_background":{"type":"boolean","description":"Run in the background and return a task id immediately (collect with task_output, stop with task_kill). No timeout applies."}},"required":["command","description"]}},{"name":"cordis_inspect","description":"Inspect the live Cordis runtime in the current DSH process. Read-only. Sections: `services` (every provided ctx service and the plugin fiber that owns it), `plugins` (all live plugin fibers with their lifecycle states), `tools` (the model-facing tools currently registered, i.e. what you can call), `temporary` (only temporary Plugins created by cordis_mount: id, name, state, provided services, awaited services, and lifetime), `api` (method signatures AND argument/return type shapes for every LIVE service — read this before writing plugin code that calls a service), `events` (every harness event with its dispatch mode and exact signature — pick listener targets here). Temporary Plugins exist only in memory, remain active across later turns, and disappear after cordis_unmount, toolset unload, or DSH restart; they are not restored automatically. The `temporary` section is a subset of `plugins`. Omit `what` to get all six sections. With `what:\"api\"` or `what:\"events\"`, pass an exact `name` to narrow to one service/event and include its original source JSDoc.","parameters":{"type":"object","properties":{"what":{"type":"string","description":"Limit the report to one section. Omit for all sections.","enum":["services","plugins","tools","temporary","api","events"]},"name":{"type":"string","description":"Exact service key or event name whose original JSDoc to include; valid only with what:\"api\" or what:\"events\"."}}}},{"name":"cordis_mount","description":"Mount a temporary Cordis Plugin in the current DSH process. This creates an in-memory runtime Plugin, not an installed or configured Plugin. It remains active across later turns until cordis_unmount, toolset unload, or DSH restart. It does not create files, install a package, change cordis.yml or personal/project config, survive restart, or automatically become permanent. To keep it, ask the Agent to implement a normal local, project, or repository Plugin through the regular development workflow. It may affect other sessions in the same process; the sandbox is not a security boundary, and injected services reach the real runtime. `code` runs now as the body of an async JavaScript function in an isolated sandbox and MUST `return` a plugin. Two forms: FUNCTION form `return (ctx) => { … }` — declares no inject, so it can register tools, listen to events, and provide services, but reaching ANY service (e.g. ctx.bash) throws; use it only when you need no services. OBJECT form `return { name?, inject: ['bash', 'llm', …], apply(ctx) { … } }` — declares dependencies, and cordis activates the plugin only after the services exist; PREFER this form. You may reach ONLY the services you list in inject: an undeclared service throws even if it exists, because an undeclared dependency would not be cleaned up if its provider is unmounted. BEFORE calling a service from your code, read cordis_inspect what:\"api\" — it lists method signatures AND the type shapes of their arguments/returns (do not guess a field's type; e.g. a bash run's stdout is an object, not a string). Inside `apply`, use the standard cordis API: `ctx.on(event, listener)` to observe events (see cordis_inspect what:\"events\"), or call `harness.registerTool(ctx, harness.defineTool({ name, description, parameters: { text: { type: 'string', required: true } }, output: { schema: { type: 'string' }, render(_args, value) { return [{ type: 'text', text: value }] } }, async execute(args) { return args.text } }))` to give yourself a new tool — it becomes callable on your NEXT step. Tool parameters: each key IS a property — { type: 'string'|'number'|'integer'|'boolean'|'null'|'object'|'array'|'json', required?: true, description?, enum?, const?, items?, properties? }; every direct DSL object declares additionalProperties: true|false, and oneOf: [schema, schema, ...] replaces type for an exact-one union. A raw JSON-Schema { type: 'object', properties, required?: […] } wrapper is also accepted with open-by-default objects. A tool's `execute` MUST return the lossless JSON value declared by `output.schema`; `output.render(args, value)` separately returns Native/model content blocks. Temporary Plugins can COMPOSE: one Plugin may `ctx.provide('name', value)` a service and another may declare `inject: ['name']` to consume it — the consumer stays pending until the provider exists and returns to pending when the provider is unmounted. Everything registered inside `apply` is cleaned up automatically by cordis_unmount. Sandbox globals: `console` (tagged `[cordis:]`, writes through to the harness terminal), `harness.defineTool`, `harness.registerTool`, `btoa`, `atob`, `TextEncoder`, `TextDecoder`. Node APIs are DISABLED — do filesystem/network/timer work through the cordis services, never Node built-ins: `require`, `setTimeout`/`setInterval`, and `fetch` throw redirect errors; `process` and `Buffer` are undefined. Instead use inject: ['fs'] + ctx.fs for files, inject: ['web'] + ctx.web for HTTP, inject: ['bash'] + ctx.bash for processes, and inject: ['timer'] + ctx.setTimeout/ctx.setInterval for timing (fiber effects, auto-cleaned when unmounted) — cordis_inspect what:\"api\" shows what THIS runtime provides. Write PLAIN JavaScript, not TypeScript (no `as`, no type annotations). Cautions: (1) waterfall events (e.g. tools/pre-execute) hand the listener a trailing `next` callback which MUST be called — returning without `next()` VETOES the call; prefer plain notification events unless you intend to intercept. (2) Never await something that only resolves after the current turn (your code runs INSIDE a tool call of that turn — it would deadlock). (3) Your `ctx` is a restricted façade: you can register tools, observe events, provide/consume services, and use timers, but framework internals (ctx.root, ctx.fiber, ctx.extend, ctx.plugin, …) are withheld. It is not a security boundary though — the services you inject (e.g. ctx.bash) reach the real runtime.","parameters":{"type":"object","properties":{"code":{"type":"string","description":"JavaScript body returning a temporary Plugin; evaluated now and saved nowhere."}},"required":["code"]}},{"name":"cordis_unmount","description":"Unmount a current-process temporary Plugin created by cordis_mount. Waits for its tools, listeners, services, timers, and other owned effects to clean up completely. Only dyn-N temporary ids are accepted; this cannot remove Loader, configured, or installed Plugins.","parameters":{"type":"object","properties":{"id":{"type":"string","description":"The temporary Plugin id returned by cordis_mount (for example \"dyn-1\"); valid only in this process and invalid after unmount or restart."}},"required":["id"]}},{"name":"edit","description":"Edit an existing UTF-8 text file by replacing literal text.","parameters":{"type":"object","properties":{"file_path":{"type":"string","description":"Path to edit, resolved by the filesystem backend."},"old_string":{"type":"string","description":"Literal text to replace. Must match exactly."},"new_string":{"type":"string","description":"Literal replacement text. Use an empty string to delete the match."},"replace_all":{"type":"boolean","description":"Replace all matches. Defaults to false; when false, old_string must appear exactly once."}},"required":["file_path","old_string","new_string"]}},{"name":"ralph","description":"Run a foreground fresh-agent Ralph loop toward one immutable objective. Use only when the direct human explicitly asks for Ralph or fresh-agent iteration. Each round opens a new child with no parent conversation or prior child session; the shared workspace is long-term memory, and only a bounded structured report crosses rounds. The call returns when a worker reports completion or a concrete blocker, or at the round limit. Ordinary long-running same-session work belongs to goal tools.","parameters":{"type":"object","properties":{"objective":{"type":"string","description":"The immutable completion objective for every fresh Ralph round."},"maxRounds":{"type":"number","description":"Optional positive safe-integer round cap, bounded by the deployment ceiling."}},"required":["objective"]}},{"name":"read","description":"Read a UTF-8 text file and return line-numbered content.","parameters":{"type":"object","properties":{"file_path":{"type":"string","description":"Path to read, resolved by the filesystem backend."},"offset":{"type":"number","description":"1-based first line to return. Defaults to 1."},"limit":{"type":"number","description":"Maximum number of lines to return. Defaults to 2000."}},"required":["file_path"]}},{"name":"run_code","description":"Execute a TypeScript program against the available tools. Write the BODY of an async function (erasable syntax only; top-level `await` and `return` work) and call tools as `await tools.name(args)` per the declarations in the system prompt. Only what you print or return comes back — curate it.","parameters":{"type":"object","properties":{"code":{"type":"string","description":"The program: the body of an async TypeScript function."},"description":{"type":"string","description":"Clear, concise description of what this program does in active voice, 5-10 words (shown in the UI). Examples: \"Count TODO markers across packages\"; \"Read failing test and its fixture\"; \"Rename config key in every cordis.yml\"."}},"required":["code","description"]}},{"name":"skill","description":"Load the full instructions for an available skill. Call this with the exact skill name from the session skill catalog before acting on a task that names or clearly matches that skill.","parameters":{"type":"object","properties":{"name":{"type":"string","description":"The exact skill name from the available skills list."}},"required":["name"]}},{"name":"subagent","description":"Delegate a self-contained task to a subagent (a separate agent that works in its own context) and return its final result. Use this to offload focused, independent work — research, a scoped implementation, an analysis — so it does not consume this conversation's context. The subagent runs to completion and you receive only its final answer, not its intermediate steps. Give it a complete, standalone prompt: it does not see this conversation. Set `run_in_background: true` to return a task id; collect with `task_output` and stop with `task_kill`.","parameters":{"type":"object","properties":{"description":{"type":"string","description":"A short (3-5 word) description of the delegated task, for display."},"prompt":{"type":"string","description":"The complete, self-contained task for the subagent. It does not share this conversation's context, so include everything it needs."},"run_in_background":{"type":"boolean","description":"Run as a background task and return its id; collect with task_output or stop with task_kill."}},"required":["description","prompt"]}},{"name":"subagent_fork","description":"Delegate a task to a subagent that inherits this conversation: a child agent seeded with all completed turns so far (it does not see the current in-flight turn), returning only its final result. Use this when the subtask builds on this conversation's context — a follow-up analysis, a review, a continuation — without consuming this conversation's context for the work itself. You receive only its final answer, not its intermediate steps. Set `run_in_background: true` to return a task id; collect with `task_output` and stop with `task_kill`.","parameters":{"type":"object","properties":{"description":{"type":"string","description":"A short (3-5 word) description of the delegated task, for display."},"prompt":{"type":"string","description":"The task for the subagent. It already sees this conversation's completed turns, so build on them freely and state only what is new."},"run_in_background":{"type":"boolean","description":"Run as a background task and return its id; collect with task_output or stop with task_kill."}},"required":["description","prompt"]}},{"name":"task_kill","description":"Request cancellation of a running background task by task id. Returns immediately; the task settles as killed once its work actually stops.","parameters":{"type":"object","properties":{"task_id":{"type":"string","description":"Task id returned by the tool that started the background work."},"reason":{"type":"string","description":"Optional short reason, recorded in the log and forwarded to the task."}},"required":["task_id"]}},{"name":"task_list","description":"List your background tasks (running and finished) with their ids, kinds, and statuses.","parameters":{"type":"object","properties":{}}},{"name":"task_output","description":"Read a background task. Stream tasks return only output since the previous read; final-output tasks return their result after settlement. Every response ends with `[status: ...]`. Reads are non-blocking unless `wait: true`, which waits up to the configured cap.","parameters":{"type":"object","properties":{"task_id":{"type":"string","description":"Task id returned by the tool that started the background work."},"wait":{"type":"boolean","description":"Block until the task reaches a terminal status or the timeout expires. A timed-out wait returns [status: running] and leaves the task alive."},"timeout_ms":{"type":"number","description":"Max wait in milliseconds (only meaningful with wait: true). Defaults to the configured wait timeout; capped by the configured maximum."}},"required":["task_id"]}},{"name":"todo_write","description":"Record and update a structured task list for the current work. Send the ENTIRE list every call — it REPLACES the previous list (there are no partial updates, no per-item edits). Use it to plan multi-step work and show progress: add one todo per concrete step before you start. Keep AT MOST ONE todo `in_progress` at a time; while work remains, exactly one active task should be `in_progress`. Mark a todo `completed` the moment it is done (do not batch completions), and allow no `in_progress` item only once all work is complete. Skip the list for trivial single-step tasks. Statuses: `pending` (not started), `in_progress` (being worked on now), `completed` (finished).","parameters":{"type":"object","properties":{"todos":{"type":"array","description":"The COMPLETE task list, replacing any previous list.","items":{"type":"object","additionalProperties":false,"properties":{"content":{"type":"string","description":"What the task is — a short imperative line."},"status":{"type":"string","description":"pending (not started) | in_progress (now) | completed (done).","enum":["pending","in_progress","completed"]}},"required":["content","status"]}}},"required":["todos"]}},{"name":"workflow","description":"Run a JavaScript workflow script that orchestrates subagents at scale. Use this for work that fans out across many independent pieces — an audit over many files, a migration, multi-angle research, adversarial verification of findings — where you write the orchestration as a script instead of delegating turn by turn.\n\nThe workflow's identity rides the `meta` parameter as JSON: required `name` (short kebab-case) and `description` strings, optional `whenToUse` string and `phases` array (`{title, detail?, provider?, model?}`). The `script` parameter is the plain JavaScript body ONLY (NOT TypeScript, and NO `export const meta` statement — meta is a parameter, not code), running with top-level await; end with `return ` — the value must be JSON-serializable and is this tool's result.\n\nScript-body hooks:\n- `agent(prompt, opts?): Promise` — run one subagent to completion. Without `opts.schema` it resolves to the child's final text; with `opts.schema` (an object-rooted JSON Schema using ONLY type/properties/required/additionalProperties/items/enum/const/oneOf — no pattern/format/numeric bounds) it resolves to the validated object. Resolves `null` when the child fails (filter with `.filter(Boolean)`). Other opts: `label` (display), `phase` (progress group), and independent `provider`/`model` LLM target overrides (either may be provided alone). Anything else (`effort`/`isolation`/`agentType`) is rejected loudly.\n- `pipeline(items, ...stages): Promise` — run each item through the stages independently with NO barrier between stages (prefer this for multi-stage work). Each stage receives `(prev, item, index)`. An ordinary stage throw drops that ITEM to `null` and skips its remaining stages.\n- `parallel(thunks): Promise` — run zero-argument functions concurrently and await ALL of them (a barrier; use only when a stage genuinely needs every prior result together). A throwing thunk resolves to `null`.\n- `phase(title)` — start a progress phase; `log(message)` — narrate progress; `args` — the tool call's `args` input, verbatim.\n\nMisused hooks (bad arguments, unknown options, unsupported schemas, tripped caps) throw errors that ALWAYS kill the script — they never dissolve into a per-item `null`.\n\nConstraints: concurrency and total-agent caps apply; no filesystem, network, timers, or Node.js APIs are provided — the agents do the work, the script only coordinates them. The run executes in the foreground: this call returns when the whole script finishes.","parameters":{"type":"object","properties":{"script":{"type":"string","description":"The plain-JS workflow script body (top-level await allowed; NO `export const meta` statement; end with `return `)."},"meta":{"type":"object","description":"The workflow identity block (plain JSON — never code).","additionalProperties":true,"properties":{"name":{"type":"string","description":"Short kebab-case workflow name."},"description":{"type":"string","description":"One-line description of what the workflow does."},"whenToUse":{"type":"string","description":"Optional guidance on when this workflow applies."},"phases":{"type":"array","description":"Optional phase declarations matched by phase() calls.","items":{"type":"object","additionalProperties":true,"properties":{"title":{"type":"string","description":"The phase title phase() calls match by exact string."},"detail":{"type":"string","description":"Optional one-line description of the phase."},"provider":{"type":"string","description":"Optional provider override this phase is expected to use."},"model":{"type":"string","description":"Optional model override this phase is expected to use."}},"required":["title"]}}},"required":["name","description"]},"args":{"type":"object","description":"Optional JSON input exposed to the script as the `args` global (wrap a bare list as a field, e.g. {\"files\": [...]}).","additionalProperties":true}},"required":["script","meta"]}},{"name":"write","description":"Create or fully replace a UTF-8 text file.","parameters":{"type":"object","properties":{"file_path":{"type":"string","description":"Path to write, resolved by the filesystem backend."},"content":{"type":"string","description":"Full UTF-8 text content to write."}},"required":["file_path","content"]}}]},"reason":"initial"}} +{"type":"request/header","seq":4,"time":1783957884701,"data":{"header":{"config":{"provider":"deepseek","model":"deepseek-v4-flash"},"system":"You are an AI agent powered by the DeepSeek Harness SDK.\n\nYou are headless-agent, a coding assistant powered by the deepseek-v4-flash model. Your working directory is {{cwd}}.\n\nVerify your work by running the code or tests. Keep answers brief and factual.\n\n\nUse the read tool — not shell commands like cat — to inspect text files. Results include line numbers. Use offset and limit to continue reading large files.\n\nUse the write tool to create files or completely replace file contents. Existing files are overwritten, so read an existing file first (the default fs-policy requires it) and prefer edit for targeted changes.\n\nUse the edit tool for targeted changes to existing UTF-8 text files. It replaces literal old_string with new_string; by default old_string must appear exactly once. If old_string appears multiple times, provide a more specific old_string or set replace_all to true. Read the file first (the default fs-policy requires it), unless you just created or edited it in this session.\n\nCheck the [exit code: N] marker on every bash result; investigate failures before moving on.\n\nTrack every background task id you start. You are notified in-session when a task finishes — do not busy-poll or sleep on one; keep working on independent steps and do not duplicate a running task's work. Before giving a final answer, collect every still-relevant task with task_output (set wait: true only when you are genuinely blocked on it), and task_kill tasks that stopped mattering.\n\nUse the workflow tool ONLY when the user explicitly asks for a workflow or for large multi-agent orchestration: you write a JavaScript script (the tool description documents the exact format) that fans work out across many subagents with phases and structured results. For one or two delegations, prefer plain subagent calls.\n\nUse the ralph tool ONLY when the direct human explicitly asks for a Ralph loop or fresh-agent iterative execution. Each Ralph round starts a fresh child with no conversation seed and uses the shared workspace as durable memory. Completion and blockers are worker reports, not independent evaluation. Use same-session goal tools for ordinary long-running objectives, and plain subagents or workflows for bounded delegation and fan-out.\n\n## Writing code for run_code\n\nPass `run_code` the body of an async TypeScript function (erasable syntax only — no `enum` or namespaces; type annotations are advisory, the code runs type-stripped). Inside the program:\n\n- Call tools as `await tools.name(args)` — quoted access for exotic names: `tools[\"my-tool\"](args)`. Every call resolves to the tool's typed canonical JSON value. Tool arguments must be lossless JSON.\n- A FAILED tool call rejects with `ToolCallError`, whose `toolName` identifies the failed tool and whose `message` is human-readable — `try/catch` it to handle and continue.\n- Independent read-only calls MAY overlap under `Promise.all` (safe calls run concurrently; mutating calls run alone, in submission order). Sequence dependent work with `await`.\n- Emit results with `return` and/or `console.log(...)`. ONLY what you print or return comes back to you — intermediate tool results never enter the conversation, so extract just what you need.\n\nThe available tools:\n\n```ts\ntype JsonValue = null | boolean | number | string | JsonValue[] | { [key: string]: JsonValue }\n\ninterface ToolArgsMap {\n /** Execute a bash command (`bash -c`) and return its stdout/stderr. Each call runs in a fresh shell: no state (cwd, variables, functions) persists between calls — pass `workdir` instead of using `cd`. Non-zero exits are reported as `[exit code: N]`. Current harness environment facts are exposed through managed `$DSH_*` variables; inspect them when needed. Commands may run under a file sandbox; a blocked file operation is reported as `[sandbox: file access denied under mode]` — a policy denial, not a bug in the command; do not retry another way. Long output is truncated to its tail; the full output is saved to a file whose path is reported when available. Set `run_in_background: true` for long-running commands: the call returns a task id immediately; read its output with `task_output` and stop it with `task_kill`. */\n bash: {\n /** The bash command to execute. */\n command: string;\n /** Clear, concise description of what this command does in active voice, 5-10 words (shown in the UI). Examples: \"ls\" → \"List files in current directory\"; \"git status\" → \"Show working tree status\"; \"npm install\" → \"Install package dependencies\". */\n description: string;\n /** Timeout in milliseconds. The executor applies its configured default and cap, and kills the command on expiry. */\n timeoutMs?: number;\n /** Working directory for this command. Defaults to the session workspace; a relative path is resolved against it. */\n workdir?: string;\n /** Run in the background and return a task id immediately (collect with task_output, stop with task_kill). No timeout applies. */\n run_in_background?: boolean;\n } & Record;\n /** Inspect the live Cordis runtime in the current DSH process. Read-only. Sections: `services` (every provided ctx service and the plugin fiber that owns it), `plugins` (all live plugin fibers with their lifecycle states), `tools` (the model-facing tools currently registered, i.e. what you can call), `temporary` (only temporary Plugins created by cordis_mount: id, name, state, provided services, awaited services, and lifetime), `api` (method signatures AND argument/return type shapes for every LIVE service — read this before writing plugin code that calls a service), `events` (every harness event with its dispatch mode and exact signature — pick listener targets here). Temporary Plugins exist only in memory, remain active across later turns, and disappear after cordis_unmount, toolset unload, or DSH restart; they are not restored automatically. The `temporary` section is a subset of `plugins`. Omit `what` to get all six sections. With `what:\"api\"` or `what:\"events\"`, pass an exact `name` to narrow to one service/event and include its original source JSDoc. */\n cordis_inspect: {\n /** Limit the report to one section. Omit for all sections. */\n what?: \"services\" | \"plugins\" | \"tools\" | \"temporary\" | \"api\" | \"events\";\n /** Exact service key or event name whose original JSDoc to include; valid only with what:\"api\" or what:\"events\". */\n name?: string;\n } & Record;\n /** Mount a temporary Cordis Plugin in the current DSH process. This creates an in-memory runtime Plugin, not an installed or configured Plugin. It remains active across later turns until cordis_unmount, toolset unload, or DSH restart. It does not create files, install a package, change cordis.yml or personal/project config, survive restart, or automatically become permanent. To keep it, ask the Agent to implement a normal local, project, or repository Plugin through the regular development workflow. It may affect other sessions in the same process; the sandbox is not a security boundary, and injected services reach the real runtime. `code` runs now as the body of an async JavaScript function in an isolated sandbox and MUST `return` a plugin. Two forms: FUNCTION form `return (ctx) => { … }` — declares no inject, so it can register tools, listen to events, and provide services, but reaching ANY service (e.g. ctx.bash) throws; use it only when you need no services. OBJECT form `return { name?, inject: ['bash', 'llm', …], apply(ctx) { … } }` — declares dependencies, and cordis activates the plugin only after the services exist; PREFER this form. You may reach ONLY the services you list in inject: an undeclared service throws even if it exists, because an undeclared dependency would not be cleaned up if its provider is unmounted. BEFORE calling a service from your code, read cordis_inspect what:\"api\" — it lists method signatures AND the type shapes of their arguments/returns (do not guess a field's type; e.g. a bash run's stdout is an object, not a string). Inside `apply`, use the standard cordis API: `ctx.on(event, listener)` to observe events (see cordis_inspect what:\"events\"), or call `harness.registerTool(ctx, harness.defineTool({ name, description, parameters: { text: { type: 'string', required: true } }, output: { schema: { type: 'string' }, render(_args, value) { return [{ type: 'text', text: value }] } }, async execute(args) { return args.text } }))` to give yourself a new tool — it becomes callable on your NEXT step. Tool parameters: each key IS a property — { type: 'string'|'number'|'integer'|'boolean'|'null'|'object'|'array'|'json', required?: true, description?, enum?, const?, items?, properties? }; every direct DSL object declares additionalProperties: true|false, and oneOf: [schema, schema, ...] replaces type for an exact-one union. A raw JSON-Schema { type: 'object', properties, required?: […] } wrapper is also accepted with open-by-default objects. A tool's `execute` MUST return the lossless JSON value declared by `output.schema`; `output.render(args, value)` separately returns Native/model content blocks. Temporary Plugins can COMPOSE: one Plugin may `ctx.provide('name', value)` a service and another may declare `inject: ['name']` to consume it — the consumer stays pending until the provider exists and returns to pending when the provider is unmounted. Everything registered inside `apply` is cleaned up automatically by cordis_unmount. Sandbox globals: `console` (tagged `[cordis:]`, writes through to the harness terminal), `harness.defineTool`, `harness.registerTool`, `btoa`, `atob`, `TextEncoder`, `TextDecoder`. Node APIs are DISABLED — do filesystem/network/timer work through the cordis services, never Node built-ins: `require`, `setTimeout`/`setInterval`, and `fetch` throw redirect errors; `process` and `Buffer` are undefined. Instead use inject: ['fs'] + ctx.fs for files, inject: ['web'] + ctx.web for HTTP, inject: ['bash'] + ctx.bash for processes, and inject: ['timer'] + ctx.setTimeout/ctx.setInterval for timing (fiber effects, auto-cleaned when unmounted) — cordis_inspect what:\"api\" shows what THIS runtime provides. Write PLAIN JavaScript, not TypeScript (no `as`, no type annotations). Cautions: (1) waterfall events (e.g. tools/pre-execute) hand the listener a trailing `next` callback which MUST be called — returning without `next()` VETOES the call; prefer plain notification events unless you intend to intercept. (2) Never await something that only resolves after the current turn (your code runs INSIDE a tool call of that turn — it would deadlock). (3) Your `ctx` is a restricted façade: you can register tools, observe events, provide/consume services, and use timers, but framework internals (ctx.root, ctx.fiber, ctx.extend, ctx.plugin, …) are withheld. It is not a security boundary though — the services you inject (e.g. ctx.bash) reach the real runtime. */\n cordis_mount: {\n /** JavaScript body returning a temporary Plugin; evaluated now and saved nowhere. */\n code: string;\n } & Record;\n /** Unmount a current-process temporary Plugin created by cordis_mount. Waits for its tools, listeners, services, timers, and other owned effects to clean up completely. Only dyn-N temporary ids are accepted; this cannot remove Loader, configured, or installed Plugins. */\n cordis_unmount: {\n /** The temporary Plugin id returned by cordis_mount (for example \"dyn-1\"); valid only in this process and invalid after unmount or restart. */\n id: string;\n } & Record;\n /** Edit an existing UTF-8 text file by replacing literal text. */\n edit: {\n /** Path to edit, resolved by the filesystem backend. */\n file_path: string;\n /** Literal text to replace. Must match exactly. */\n old_string: string;\n /** Literal replacement text. Use an empty string to delete the match. */\n new_string: string;\n /** Replace all matches. Defaults to false; when false, old_string must appear exactly once. */\n replace_all?: boolean;\n } & Record;\n /** Run a foreground fresh-agent Ralph loop toward one immutable objective. Use only when the direct human explicitly asks for Ralph or fresh-agent iteration. Each round opens a new child with no parent conversation or prior child session; the shared workspace is long-term memory, and only a bounded structured report crosses rounds. The call returns when a worker reports completion or a concrete blocker, or at the round limit. Ordinary long-running same-session work belongs to goal tools. */\n ralph: {\n /** The immutable completion objective for every fresh Ralph round. */\n objective: string;\n /** Optional positive safe-integer round cap, bounded by the deployment ceiling. */\n maxRounds?: number;\n } & Record;\n /** Read a UTF-8 text file and return line-numbered content. */\n read: {\n /** Path to read, resolved by the filesystem backend. */\n file_path: string;\n /** 1-based first line to return. Defaults to 1. */\n offset?: number;\n /** Maximum number of lines to return. Defaults to 2000. */\n limit?: number;\n } & Record;\n /** Load the full instructions for an available skill. Call this with the exact skill name from the session skill catalog before acting on a task that names or clearly matches that skill. */\n skill: {\n /** The exact skill name from the available skills list. */\n name: string;\n } & Record;\n /** Delegate a self-contained task to a subagent (a separate agent that works in its own context) and return its final result. Use this to offload focused, independent work — research, a scoped implementation, an analysis — so it does not consume this conversation's context. The subagent runs to completion and you receive only its final answer, not its intermediate steps. Give it a complete, standalone prompt: it does not see this conversation. Set `run_in_background: true` to return a task id; collect with `task_output` and stop with `task_kill`. */\n subagent: {\n /** A short (3-5 word) description of the delegated task, for display. */\n description: string;\n /** The complete, self-contained task for the subagent. It does not share this conversation's context, so include everything it needs. */\n prompt: string;\n /** Run as a background task and return its id; collect with task_output or stop with task_kill. */\n run_in_background?: boolean;\n } & Record;\n /** Delegate a task to a subagent that inherits this conversation: a child agent seeded with all completed turns so far (it does not see the current in-flight turn), returning only its final result. Use this when the subtask builds on this conversation's context — a follow-up analysis, a review, a continuation — without consuming this conversation's context for the work itself. You receive only its final answer, not its intermediate steps. Set `run_in_background: true` to return a task id; collect with `task_output` and stop with `task_kill`. */\n subagent_fork: {\n /** A short (3-5 word) description of the delegated task, for display. */\n description: string;\n /** The task for the subagent. It already sees this conversation's completed turns, so build on them freely and state only what is new. */\n prompt: string;\n /** Run as a background task and return its id; collect with task_output or stop with task_kill. */\n run_in_background?: boolean;\n } & Record;\n /** Request cancellation of a running background task by task id. Returns immediately; the task settles as killed once its work actually stops. */\n task_kill: {\n /** Task id returned by the tool that started the background work. */\n task_id: string;\n /** Optional short reason, recorded in the log and forwarded to the task. */\n reason?: string;\n } & Record;\n /** List your background tasks (running and finished) with their ids, kinds, and statuses. */\n task_list: Record;\n /** Read a background task. Stream tasks return only output since the previous read; final-output tasks return their result after settlement. Every response ends with `[status: ...]`. Reads are non-blocking unless `wait: true`, which waits up to the configured cap. */\n task_output: {\n /** Task id returned by the tool that started the background work. */\n task_id: string;\n /** Block until the task reaches a terminal status or the timeout expires. A timed-out wait returns [status: running] and leaves the task alive. */\n wait?: boolean;\n /** Max wait in milliseconds (only meaningful with wait: true). Defaults to the configured wait timeout; capped by the configured maximum. */\n timeout_ms?: number;\n } & Record;\n /** Record and update a structured task list for the current work. Send the ENTIRE list every call — it REPLACES the previous list (there are no partial updates, no per-item edits). Use it to plan multi-step work and show progress: add one todo per concrete step before you start. Keep AT MOST ONE todo `in_progress` at a time; while work remains, exactly one active task should be `in_progress`. Mark a todo `completed` the moment it is done (do not batch completions), and allow no `in_progress` item only once all work is complete. Skip the list for trivial single-step tasks. Statuses: `pending` (not started), `in_progress` (being worked on now), `completed` (finished). */\n todo_write: {\n /** The COMPLETE task list, replacing any previous list. */\n todos: ({\n /** What the task is — a short imperative line. */\n content: string;\n /** pending (not started) | in_progress (now) | completed (done). */\n status: \"pending\" | \"in_progress\" | \"completed\";\n })[];\n } & Record;\n /** Run a JavaScript workflow script that orchestrates subagents at scale. Use this for work that fans out across many independent pieces — an audit over many files, a migration, multi-angle research, adversarial verification of findings — where you write the orchestration as a script instead of delegating turn by turn. The workflow's identity rides the `meta` parameter as JSON: required `name` (short kebab-case) and `description` strings, optional `whenToUse` string and `phases` array (`{title, detail?, provider?, model?}`). The `script` parameter is the plain JavaScript body ONLY (NOT TypeScript, and NO `export const meta` statement — meta is a parameter, not code), running with top-level await; end with `return ` — the value must be JSON-serializable and is this tool's result. Script-body hooks: - `agent(prompt, opts?): Promise` — run one subagent to completion. Without `opts.schema` it resolves to the child's final text; with `opts.schema` (an object-rooted JSON Schema using ONLY type/properties/required/additionalProperties/items/enum/const/oneOf — no pattern/format/numeric bounds) it resolves to the validated object. Resolves `null` when the child fails (filter with `.filter(Boolean)`). Other opts: `label` (display), `phase` (progress group), and independent `provider`/`model` LLM target overrides (either may be provided alone). Anything else (`effort`/`isolation`/`agentType`) is rejected loudly. - `pipeline(items, ...stages): Promise` — run each item through the stages independently with NO barrier between stages (prefer this for multi-stage work). Each stage receives `(prev, item, index)`. An ordinary stage throw drops that ITEM to `null` and skips its remaining stages. - `parallel(thunks): Promise` — run zero-argument functions concurrently and await ALL of them (a barrier; use only when a stage genuinely needs every prior result together). A throwing thunk resolves to `null`. - `phase(title)` — start a progress phase; `log(message)` — narrate progress; `args` — the tool call's `args` input, verbatim. Misused hooks (bad arguments, unknown options, unsupported schemas, tripped caps) throw errors that ALWAYS kill the script — they never dissolve into a per-item `null`. Constraints: concurrency and total-agent caps apply; no filesystem, network, timers, or Node.js APIs are provided — the agents do the work, the script only coordinates them. The run executes in the foreground: this call returns when the whole script finishes. */\n workflow: {\n /** The plain-JS workflow script body (top-level await allowed; NO `export const meta` statement; end with `return `). */\n script: string;\n /** The workflow identity block (plain JSON — never code). */\n meta: {\n /** Short kebab-case workflow name. */\n name: string;\n /** One-line description of what the workflow does. */\n description: string;\n /** Optional guidance on when this workflow applies. */\n whenToUse?: string;\n /** Optional phase declarations matched by phase() calls. */\n phases?: ({\n /** The phase title phase() calls match by exact string. */\n title: string;\n /** Optional one-line description of the phase. */\n detail?: string;\n /** Optional provider override this phase is expected to use. */\n provider?: string;\n /** Optional model override this phase is expected to use. */\n model?: string;\n } & Record)[];\n } & Record;\n /** Optional JSON input exposed to the script as the `args` global (wrap a bare list as a field, e.g. {\"files\": [...]}). */\n args?: Record;\n } & Record;\n /** Create or fully replace a UTF-8 text file. */\n write: {\n /** Path to write, resolved by the filesystem backend. */\n file_path: string;\n /** Full UTF-8 text content to write. */\n content: string;\n } & Record;\n}\n\ninterface ToolOutputMap {\n bash: {\n kind: \"background\";\n taskId: string;\n } | {\n kind: \"foreground\";\n exitCode: number | null;\n signal: string | null;\n timedOut: boolean;\n aborted: boolean;\n timeoutMs: number;\n stdout: {\n text: string;\n truncated: boolean;\n spillPath?: string;\n };\n stderr: {\n text: string;\n truncated: boolean;\n spillPath?: string;\n };\n sandbox?: {\n mode: string;\n denied: boolean;\n enforcement?: string;\n runnerFailed?: boolean;\n };\n };\n cordis_inspect: string;\n cordis_mount: {\n id: string;\n pluginName: string;\n state: \"pending\" | \"loading\" | \"active\" | \"failed\" | \"disposed\" | \"unloading\";\n provides: string[];\n waitingFor: string[];\n };\n cordis_unmount: {\n id: string;\n pluginName: string;\n };\n edit: {\n path: string;\n before: string;\n after: string;\n };\n ralph: {\n runId: string;\n agentsStarted: number;\n result: JsonValue;\n };\n read: {\n path: string;\n offset: number;\n lines: {\n number: number;\n text: string;\n }[];\n totalLines: number;\n };\n skill: {\n name: string;\n provider: string;\n resourceBase?: {\n kind: \"directory\";\n path: string;\n } | {\n kind: \"url\";\n url: string;\n } | {\n kind: \"opaque\";\n description: string;\n };\n content: string;\n };\n subagent: {\n kind: \"background\";\n taskId: string;\n } | {\n kind: \"foreground\";\n runId: string;\n output: JsonValue[];\n };\n subagent_fork: {\n kind: \"background\";\n taskId: string;\n } | {\n kind: \"foreground\";\n runId: string;\n output: JsonValue[];\n };\n task_kill: {\n outcome: \"cancellation-requested\" | \"already-finished\";\n task: {\n id: string;\n kind: string;\n label: string;\n status: \"running\" | \"stopping\" | \"completed\" | \"killed\" | \"failed\";\n detail?: string;\n startedAt: number;\n finishedAt?: number;\n };\n };\n task_list: ({\n id: string;\n kind: string;\n label: string;\n status: \"running\" | \"stopping\" | \"completed\" | \"killed\" | \"failed\";\n detail?: string;\n startedAt: number;\n finishedAt?: number;\n })[];\n task_output: {\n text: string;\n task: {\n id: string;\n kind: string;\n label: string;\n status: \"running\" | \"stopping\" | \"completed\" | \"killed\" | \"failed\";\n detail?: string;\n startedAt: number;\n finishedAt?: number;\n };\n };\n todo_write: {\n todos: ({\n content: string;\n status: \"pending\" | \"in_progress\" | \"completed\";\n })[];\n counts: {\n pending: number;\n inProgress: number;\n completed: number;\n };\n };\n workflow: {\n runId: string;\n agentsStarted: number;\n result: JsonValue;\n };\n write: {\n path: string;\n operation: \"create\" | \"update\";\n before: string | null;\n after: string;\n };\n}\n\ntype ToolName = keyof ToolOutputMap\n\ndeclare class ToolCallError extends Error {\n readonly name: \"ToolCallError\";\n readonly toolName: ToolName;\n}\n\ndeclare const tools: {\n [K in ToolName]: (args: ToolArgsMap[K]) => Promise;\n}\n```","tools":[{"name":"bash","description":"Execute a bash command (`bash -c`) and return its stdout/stderr. Each call runs in a fresh shell: no state (cwd, variables, functions) persists between calls — pass `workdir` instead of using `cd`. Non-zero exits are reported as `[exit code: N]`. Current harness environment facts are exposed through managed `$DSH_*` variables; inspect them when needed. Commands may run under a file sandbox; a blocked file operation is reported as `[sandbox: file access denied under mode]` — a policy denial, not a bug in the command; do not retry another way. Long output is truncated to its tail; the full output is saved to a file whose path is reported when available. Set `run_in_background: true` for long-running commands: the call returns a task id immediately; read its output with `task_output` and stop it with `task_kill`.","parameters":{"type":"object","properties":{"command":{"type":"string","description":"The bash command to execute."},"description":{"type":"string","description":"Clear, concise description of what this command does in active voice, 5-10 words (shown in the UI). Examples: \"ls\" → \"List files in current directory\"; \"git status\" → \"Show working tree status\"; \"npm install\" → \"Install package dependencies\"."},"timeoutMs":{"type":"number","description":"Timeout in milliseconds. The executor applies its configured default and cap, and kills the command on expiry."},"workdir":{"type":"string","description":"Working directory for this command. Defaults to the session workspace; a relative path is resolved against it."},"run_in_background":{"type":"boolean","description":"Run in the background and return a task id immediately (collect with task_output, stop with task_kill). No timeout applies."}},"required":["command","description"]}},{"name":"cordis_inspect","description":"Inspect the live Cordis runtime in the current DSH process. Read-only. Sections: `services` (every provided ctx service and the plugin fiber that owns it), `plugins` (all live plugin fibers with their lifecycle states), `tools` (the model-facing tools currently registered, i.e. what you can call), `temporary` (only temporary Plugins created by cordis_mount: id, name, state, provided services, awaited services, and lifetime), `api` (method signatures AND argument/return type shapes for every LIVE service — read this before writing plugin code that calls a service), `events` (every harness event with its dispatch mode and exact signature — pick listener targets here). Temporary Plugins exist only in memory, remain active across later turns, and disappear after cordis_unmount, toolset unload, or DSH restart; they are not restored automatically. The `temporary` section is a subset of `plugins`. Omit `what` to get all six sections. With `what:\"api\"` or `what:\"events\"`, pass an exact `name` to narrow to one service/event and include its original source JSDoc.","parameters":{"type":"object","properties":{"what":{"type":"string","description":"Limit the report to one section. Omit for all sections.","enum":["services","plugins","tools","temporary","api","events"]},"name":{"type":"string","description":"Exact service key or event name whose original JSDoc to include; valid only with what:\"api\" or what:\"events\"."}}}},{"name":"cordis_mount","description":"Mount a temporary Cordis Plugin in the current DSH process. This creates an in-memory runtime Plugin, not an installed or configured Plugin. It remains active across later turns until cordis_unmount, toolset unload, or DSH restart. It does not create files, install a package, change cordis.yml or personal/project config, survive restart, or automatically become permanent. To keep it, ask the Agent to implement a normal local, project, or repository Plugin through the regular development workflow. It may affect other sessions in the same process; the sandbox is not a security boundary, and injected services reach the real runtime. `code` runs now as the body of an async JavaScript function in an isolated sandbox and MUST `return` a plugin. Two forms: FUNCTION form `return (ctx) => { … }` — declares no inject, so it can register tools, listen to events, and provide services, but reaching ANY service (e.g. ctx.bash) throws; use it only when you need no services. OBJECT form `return { name?, inject: ['bash', 'llm', …], apply(ctx) { … } }` — declares dependencies, and cordis activates the plugin only after the services exist; PREFER this form. You may reach ONLY the services you list in inject: an undeclared service throws even if it exists, because an undeclared dependency would not be cleaned up if its provider is unmounted. BEFORE calling a service from your code, read cordis_inspect what:\"api\" — it lists method signatures AND the type shapes of their arguments/returns (do not guess a field's type; e.g. a bash run's stdout is an object, not a string). Inside `apply`, use the standard cordis API: `ctx.on(event, listener)` to observe events (see cordis_inspect what:\"events\"), or call `harness.registerTool(ctx, harness.defineTool({ name, description, parameters: { text: { type: 'string', required: true } }, output: { schema: { type: 'string' }, render(_args, value) { return [{ type: 'text', text: value }] } }, async execute(args) { return args.text } }))` to give yourself a new tool — it becomes callable on your NEXT step. Tool parameters: each key IS a property — { type: 'string'|'number'|'integer'|'boolean'|'null'|'object'|'array'|'json', required?: true, description?, enum?, const?, items?, properties? }; every direct DSL object declares additionalProperties: true|false, and oneOf: [schema, schema, ...] replaces type for an exact-one union. A raw JSON-Schema { type: 'object', properties, required?: […] } wrapper is also accepted with open-by-default objects. A tool's `execute` MUST return the lossless JSON value declared by `output.schema`; `output.render(args, value)` separately returns Native/model content blocks. Temporary Plugins can COMPOSE: one Plugin may `ctx.provide('name', value)` a service and another may declare `inject: ['name']` to consume it — the consumer stays pending until the provider exists and returns to pending when the provider is unmounted. Everything registered inside `apply` is cleaned up automatically by cordis_unmount. Sandbox globals: `console` (tagged `[cordis:]`, writes through to the harness terminal), `harness.defineTool`, `harness.registerTool`, `btoa`, `atob`, `TextEncoder`, `TextDecoder`. Node APIs are DISABLED — do filesystem/network/timer work through the cordis services, never Node built-ins: `require`, `setTimeout`/`setInterval`, and `fetch` throw redirect errors; `process` and `Buffer` are undefined. Instead use inject: ['fs'] + ctx.fs for files, inject: ['web'] + ctx.web for HTTP, inject: ['bash'] + ctx.bash for processes, and inject: ['timer'] + ctx.setTimeout/ctx.setInterval for timing (fiber effects, auto-cleaned when unmounted) — cordis_inspect what:\"api\" shows what THIS runtime provides. Write PLAIN JavaScript, not TypeScript (no `as`, no type annotations). Cautions: (1) waterfall events (e.g. tools/pre-execute) hand the listener a trailing `next` callback which MUST be called — returning without `next()` VETOES the call; prefer plain notification events unless you intend to intercept. (2) Never await something that only resolves after the current turn (your code runs INSIDE a tool call of that turn — it would deadlock). (3) Your `ctx` is a restricted façade: you can register tools, observe events, provide/consume services, and use timers, but framework internals (ctx.root, ctx.fiber, ctx.extend, ctx.plugin, …) are withheld. It is not a security boundary though — the services you inject (e.g. ctx.bash) reach the real runtime.","parameters":{"type":"object","properties":{"code":{"type":"string","description":"JavaScript body returning a temporary Plugin; evaluated now and saved nowhere."}},"required":["code"]}},{"name":"cordis_unmount","description":"Unmount a current-process temporary Plugin created by cordis_mount. Waits for its tools, listeners, services, timers, and other owned effects to clean up completely. Only dyn-N temporary ids are accepted; this cannot remove Loader, configured, or installed Plugins.","parameters":{"type":"object","properties":{"id":{"type":"string","description":"The temporary Plugin id returned by cordis_mount (for example \"dyn-1\"); valid only in this process and invalid after unmount or restart."}},"required":["id"]}},{"name":"edit","description":"Edit an existing UTF-8 text file by replacing literal text.","parameters":{"type":"object","properties":{"file_path":{"type":"string","description":"Path to edit, resolved by the filesystem backend."},"old_string":{"type":"string","description":"Literal text to replace. Must match exactly."},"new_string":{"type":"string","description":"Literal replacement text. Use an empty string to delete the match."},"replace_all":{"type":"boolean","description":"Replace all matches. Defaults to false; when false, old_string must appear exactly once."}},"required":["file_path","old_string","new_string"]}},{"name":"ralph","description":"Run a foreground fresh-agent Ralph loop toward one immutable objective. Use only when the direct human explicitly asks for Ralph or fresh-agent iteration. Each round opens a new child with no parent conversation or prior child session; the shared workspace is long-term memory, and only a bounded structured report crosses rounds. The call returns when a worker reports completion or a concrete blocker, or at the round limit. Ordinary long-running same-session work belongs to goal tools.","parameters":{"type":"object","properties":{"objective":{"type":"string","description":"The immutable completion objective for every fresh Ralph round."},"maxRounds":{"type":"number","description":"Optional positive safe-integer round cap, bounded by the deployment ceiling."}},"required":["objective"]}},{"name":"read","description":"Read a UTF-8 text file and return line-numbered content.","parameters":{"type":"object","properties":{"file_path":{"type":"string","description":"Path to read, resolved by the filesystem backend."},"offset":{"type":"number","description":"1-based first line to return. Defaults to 1."},"limit":{"type":"number","description":"Maximum number of lines to return. Defaults to 2000."}},"required":["file_path"]}},{"name":"run_code","description":"Execute a TypeScript program against the available tools. Write the BODY of an async function (erasable syntax only; top-level `await` and `return` work) and call tools as `await tools.name(args)` per the declarations in the system prompt. Only what you print or return comes back — curate it.","parameters":{"type":"object","properties":{"code":{"type":"string","description":"The program: the body of an async TypeScript function."},"description":{"type":"string","description":"Clear, concise description of what this program does in active voice, 5-10 words (shown in the UI). Examples: \"Count TODO markers across packages\"; \"Read failing test and its fixture\"; \"Rename config key in every cordis.yml\"."}},"required":["code","description"]}},{"name":"skill","description":"Load the full instructions for an available skill. Call this with the exact skill name from the session skill catalog before acting on a task that names or clearly matches that skill.","parameters":{"type":"object","properties":{"name":{"type":"string","description":"The exact skill name from the available skills list."}},"required":["name"]}},{"name":"subagent","description":"Delegate a self-contained task to a subagent (a separate agent that works in its own context) and return its final result. Use this to offload focused, independent work — research, a scoped implementation, an analysis — so it does not consume this conversation's context. The subagent runs to completion and you receive only its final answer, not its intermediate steps. Give it a complete, standalone prompt: it does not see this conversation. Set `run_in_background: true` to return a task id; collect with `task_output` and stop with `task_kill`.","parameters":{"type":"object","properties":{"description":{"type":"string","description":"A short (3-5 word) description of the delegated task, for display."},"prompt":{"type":"string","description":"The complete, self-contained task for the subagent. It does not share this conversation's context, so include everything it needs."},"run_in_background":{"type":"boolean","description":"Run as a background task and return its id; collect with task_output or stop with task_kill."}},"required":["description","prompt"]}},{"name":"subagent_fork","description":"Delegate a task to a subagent that inherits this conversation: a child agent seeded with all completed turns so far (it does not see the current in-flight turn), returning only its final result. Use this when the subtask builds on this conversation's context — a follow-up analysis, a review, a continuation — without consuming this conversation's context for the work itself. You receive only its final answer, not its intermediate steps. Set `run_in_background: true` to return a task id; collect with `task_output` and stop with `task_kill`.","parameters":{"type":"object","properties":{"description":{"type":"string","description":"A short (3-5 word) description of the delegated task, for display."},"prompt":{"type":"string","description":"The task for the subagent. It already sees this conversation's completed turns, so build on them freely and state only what is new."},"run_in_background":{"type":"boolean","description":"Run as a background task and return its id; collect with task_output or stop with task_kill."}},"required":["description","prompt"]}},{"name":"task_kill","description":"Request cancellation of a running background task by task id. Returns immediately; the task settles as killed once its work actually stops.","parameters":{"type":"object","properties":{"task_id":{"type":"string","description":"Task id returned by the tool that started the background work."},"reason":{"type":"string","description":"Optional short reason, recorded in the log and forwarded to the task."}},"required":["task_id"]}},{"name":"task_list","description":"List your background tasks (running and finished) with their ids, kinds, and statuses.","parameters":{"type":"object","properties":{}}},{"name":"task_output","description":"Read a background task. Stream tasks return only output since the previous read; final-output tasks return their result after settlement. Every response ends with `[status: ...]`. Reads are non-blocking unless `wait: true`, which waits up to the configured cap.","parameters":{"type":"object","properties":{"task_id":{"type":"string","description":"Task id returned by the tool that started the background work."},"wait":{"type":"boolean","description":"Block until the task reaches a terminal status or the timeout expires. A timed-out wait returns [status: running] and leaves the task alive."},"timeout_ms":{"type":"number","description":"Max wait in milliseconds (only meaningful with wait: true). Defaults to the configured wait timeout; capped by the configured maximum."}},"required":["task_id"]}},{"name":"todo_write","description":"Record and update a structured task list for the current work. Send the ENTIRE list every call — it REPLACES the previous list (there are no partial updates, no per-item edits). Use it to plan multi-step work and show progress: add one todo per concrete step before you start. Keep AT MOST ONE todo `in_progress` at a time; while work remains, exactly one active task should be `in_progress`. Mark a todo `completed` the moment it is done (do not batch completions), and allow no `in_progress` item only once all work is complete. Skip the list for trivial single-step tasks. Statuses: `pending` (not started), `in_progress` (being worked on now), `completed` (finished).","parameters":{"type":"object","properties":{"todos":{"type":"array","description":"The COMPLETE task list, replacing any previous list.","items":{"type":"object","additionalProperties":false,"properties":{"content":{"type":"string","description":"What the task is — a short imperative line."},"status":{"type":"string","description":"pending (not started) | in_progress (now) | completed (done).","enum":["pending","in_progress","completed"]}},"required":["content","status"]}}},"required":["todos"]}},{"name":"workflow","description":"Run a JavaScript workflow script that orchestrates subagents at scale. Use this for work that fans out across many independent pieces — an audit over many files, a migration, multi-angle research, adversarial verification of findings — where you write the orchestration as a script instead of delegating turn by turn.\n\nThe workflow's identity rides the `meta` parameter as JSON: required `name` (short kebab-case) and `description` strings, optional `whenToUse` string and `phases` array (`{title, detail?, provider?, model?}`). The `script` parameter is the plain JavaScript body ONLY (NOT TypeScript, and NO `export const meta` statement — meta is a parameter, not code), running with top-level await; end with `return ` — the value must be JSON-serializable and is this tool's result.\n\nScript-body hooks:\n- `agent(prompt, opts?): Promise` — run one subagent to completion. Without `opts.schema` it resolves to the child's final text; with `opts.schema` (an object-rooted JSON Schema using ONLY type/properties/required/additionalProperties/items/enum/const/oneOf — no pattern/format/numeric bounds) it resolves to the validated object. Resolves `null` when the child fails (filter with `.filter(Boolean)`). Other opts: `label` (display), `phase` (progress group), and independent `provider`/`model` LLM target overrides (either may be provided alone). Anything else (`effort`/`isolation`/`agentType`) is rejected loudly.\n- `pipeline(items, ...stages): Promise` — run each item through the stages independently with NO barrier between stages (prefer this for multi-stage work). Each stage receives `(prev, item, index)`. An ordinary stage throw drops that ITEM to `null` and skips its remaining stages.\n- `parallel(thunks): Promise` — run zero-argument functions concurrently and await ALL of them (a barrier; use only when a stage genuinely needs every prior result together). A throwing thunk resolves to `null`.\n- `phase(title)` — start a progress phase; `log(message)` — narrate progress; `args` — the tool call's `args` input, verbatim.\n\nMisused hooks (bad arguments, unknown options, unsupported schemas, tripped caps) throw errors that ALWAYS kill the script — they never dissolve into a per-item `null`.\n\nConstraints: concurrency and total-agent caps apply; no filesystem, network, timers, or Node.js APIs are provided — the agents do the work, the script only coordinates them. The run executes in the foreground: this call returns when the whole script finishes.","parameters":{"type":"object","properties":{"script":{"type":"string","description":"The plain-JS workflow script body (top-level await allowed; NO `export const meta` statement; end with `return `)."},"meta":{"type":"object","description":"The workflow identity block (plain JSON — never code).","additionalProperties":true,"properties":{"name":{"type":"string","description":"Short kebab-case workflow name."},"description":{"type":"string","description":"One-line description of what the workflow does."},"whenToUse":{"type":"string","description":"Optional guidance on when this workflow applies."},"phases":{"type":"array","description":"Optional phase declarations matched by phase() calls.","items":{"type":"object","additionalProperties":true,"properties":{"title":{"type":"string","description":"The phase title phase() calls match by exact string."},"detail":{"type":"string","description":"Optional one-line description of the phase."},"provider":{"type":"string","description":"Optional provider override this phase is expected to use."},"model":{"type":"string","description":"Optional model override this phase is expected to use."}},"required":["title"]}}},"required":["name","description"]},"args":{"type":"object","description":"Optional JSON input exposed to the script as the `args` global (wrap a bare list as a field, e.g. {\"files\": [...]}).","additionalProperties":true}},"required":["script","meta"]}},{"name":"write","description":"Create or fully replace a UTF-8 text file.","parameters":{"type":"object","properties":{"file_path":{"type":"string","description":"Path to write, resolved by the filesystem backend."},"content":{"type":"string","description":"Full UTF-8 text content to write."}},"required":["file_path","content"]}}]},"reason":"initial"}} {"type":"assistant/chunk","seq":5,"time":1783950002005,"data":{"turn":1,"step":1,"chunk":{"type":"block-start","index":0,"blockType":"text"}}} {"type":"assistant/chunk","seq":6,"time":1783957884701,"data":{"turn":1,"step":1,"chunk":{"type":"text-delta","index":0,"text":"WORKFLOW_CHILD_OK"}}} {"type":"assistant/chunk","seq":7,"time":1783957884701,"data":{"turn":1,"step":1,"chunk":{"type":"block-end","index":0,"block":{"type":"text","text":"WORKFLOW_CHILD_OK"}}}} diff --git a/examples/headless-agent/tests/snapshots/advanced-toolchain/session.jsonl b/examples/headless-agent/tests/snapshots/advanced-toolchain/session.jsonl index 2ec15fd69d..c0d6ce1b4b 100644 --- a/examples/headless-agent/tests/snapshots/advanced-toolchain/session.jsonl +++ b/examples/headless-agent/tests/snapshots/advanced-toolchain/session.jsonl @@ -1,9 +1,9 @@ -{"type":"session","version":0,"id":"11111111-1111-4111-8111-111111111111","createdAt":1783950000000,"cwd":"/tmp/advanced-headless","delegationDepth":0} +{"type":"session","version":0,"id":"11111111-1111-4111-8111-111111111111","createdAt":1783950000000,"cwd":"{{cwd}}","delegationDepth":0} {"type":"turn/start","seq":0,"time":1783957884479,"data":{"turn":1,"trigger":{"kind":"message","source":{"kind":"user"}}}} {"type":"user/message","seq":1,"time":1783957884479,"data":{"content":[{"type":"text","text":"Run this advanced flow exactly once: try a no-op temporary Cordis Plugin named snapshot-marker; use run_code to inspect the live temporary Plugins through tools.cordis_inspect; delegate once to a direct spawn child; run one workflow that delegates to another spawn child; stop dyn-1; then reply with exactly ADVANCED_HEADLESS_OK."}],"source":{"kind":"user"},"role":"user","id":"0dda35fe-e148-4400-b837-2f6e6fe40ae6"},"surfaceOp":"append"} {"type":"session/title","seq":2,"time":1783957884479,"data":{"title":"Run this advanced flow exactly","messageSeqs":[1],"source":{"kind":"fallback"}}} {"type":"step/start","seq":3,"time":1783957884486,"data":{"turn":1,"step":1}} -{"type":"request/header","seq":4,"time":1783957884486,"data":{"header":{"config":{"provider":"deepseek","model":"deepseek-v4-flash"},"system":"You are an AI agent powered by the DeepSeek Harness SDK.\n\nYou are headless-agent, a coding assistant powered by the deepseek-v4-flash model. Your working directory is /tmp/advanced-headless.\n\nVerify your work by running the code or tests. Keep answers brief and factual.\n\n\nUse the read tool — not shell commands like cat — to inspect text files. Results include line numbers. Use offset and limit to continue reading large files.\n\nUse the write tool to create files or completely replace file contents. Existing files are overwritten, so read an existing file first (the default fs-policy requires it) and prefer edit for targeted changes.\n\nUse the edit tool for targeted changes to existing UTF-8 text files. It replaces literal old_string with new_string; by default old_string must appear exactly once. If old_string appears multiple times, provide a more specific old_string or set replace_all to true. Read the file first (the default fs-policy requires it), unless you just created or edited it in this session.\n\nCheck the [exit code: N] marker on every bash result; investigate failures before moving on.\n\nTrack every background task id you start. You are notified in-session when a task finishes — do not busy-poll or sleep on one; keep working on independent steps and do not duplicate a running task's work. Before giving a final answer, collect every still-relevant task with task_output (set wait: true only when you are genuinely blocked on it), and task_kill tasks that stopped mattering.\n\nUse the workflow tool ONLY when the user explicitly asks for a workflow or for large multi-agent orchestration: you write a JavaScript script (the tool description documents the exact format) that fans work out across many subagents with phases and structured results. For one or two delegations, prefer plain subagent calls.\n\nUse the ralph tool ONLY when the direct human explicitly asks for a Ralph loop or fresh-agent iterative execution. Each Ralph round starts a fresh child with no conversation seed and uses the shared workspace as durable memory. Completion and blockers are worker reports, not independent evaluation. Use same-session goal tools for ordinary long-running objectives, and plain subagents or workflows for bounded delegation and fan-out.\n\n## Writing code for run_code\n\nPass `run_code` the body of an async TypeScript function (erasable syntax only — no `enum` or namespaces; type annotations are advisory, the code runs type-stripped). Inside the program:\n\n- Call tools as `await tools.name(args)` — quoted access for exotic names: `tools[\"my-tool\"](args)`. Every call resolves to the tool's typed canonical JSON value. Tool arguments must be lossless JSON.\n- A FAILED tool call rejects with `ToolCallError`, whose `toolName` identifies the failed tool and whose `message` is human-readable — `try/catch` it to handle and continue.\n- Independent read-only calls MAY overlap under `Promise.all` (safe calls run concurrently; mutating calls run alone, in submission order). Sequence dependent work with `await`.\n- Emit results with `return` and/or `console.log(...)`. ONLY what you print or return comes back to you — intermediate tool results never enter the conversation, so extract just what you need.\n\nThe available tools:\n\n```ts\ntype JsonValue = null | boolean | number | string | JsonValue[] | { [key: string]: JsonValue }\n\ninterface ToolArgsMap {\n /** Execute a bash command (`bash -c`) and return its stdout/stderr. Each call runs in a fresh shell: no state (cwd, variables, functions) persists between calls — pass `workdir` instead of using `cd`. Non-zero exits are reported as `[exit code: N]`. Current harness environment facts are exposed through managed `$DSH_*` variables; inspect them when needed. Commands may run under a file sandbox; a blocked file operation is reported as `[sandbox: file access denied under mode]` — a policy denial, not a bug in the command; do not retry another way. Long output is truncated to its tail; the full output is saved to a file whose path is reported when available. Set `run_in_background: true` for long-running commands: the call returns a task id immediately; read its output with `task_output` and stop it with `task_kill`. */\n bash: {\n /** The bash command to execute. */\n command: string;\n /** Clear, concise description of what this command does in active voice, 5-10 words (shown in the UI). Examples: \"ls\" → \"List files in current directory\"; \"git status\" → \"Show working tree status\"; \"npm install\" → \"Install package dependencies\". */\n description: string;\n /** Timeout in milliseconds. The executor applies its configured default and cap, and kills the command on expiry. */\n timeoutMs?: number;\n /** Working directory for this command. Defaults to the session workspace; a relative path is resolved against it. */\n workdir?: string;\n /** Run in the background and return a task id immediately (collect with task_output, stop with task_kill). No timeout applies. */\n run_in_background?: boolean;\n } & Record;\n /** Inspect the live Cordis runtime in the current DSH process. Read-only. Sections: `services` (every provided ctx service and the plugin fiber that owns it), `plugins` (all live plugin fibers with their lifecycle states), `tools` (the model-facing tools currently registered, i.e. what you can call), `temporary` (only temporary Plugins created by cordis_mount: id, name, state, provided services, awaited services, and lifetime), `api` (method signatures AND argument/return type shapes for every LIVE service — read this before writing plugin code that calls a service), `events` (every harness event with its dispatch mode and exact signature — pick listener targets here). Temporary Plugins exist only in memory, remain active across later turns, and disappear after cordis_unmount, toolset unload, or DSH restart; they are not restored automatically. The `temporary` section is a subset of `plugins`. Omit `what` to get all six sections. With `what:\"api\"` or `what:\"events\"`, pass an exact `name` to narrow to one service/event and include its original source JSDoc. */\n cordis_inspect: {\n /** Limit the report to one section. Omit for all sections. */\n what?: \"services\" | \"plugins\" | \"tools\" | \"temporary\" | \"api\" | \"events\";\n /** Exact service key or event name whose original JSDoc to include; valid only with what:\"api\" or what:\"events\". */\n name?: string;\n } & Record;\n /** Mount a temporary Cordis Plugin in the current DSH process. This creates an in-memory runtime Plugin, not an installed or configured Plugin. It remains active across later turns until cordis_unmount, toolset unload, or DSH restart. It does not create files, install a package, change cordis.yml or personal/project config, survive restart, or automatically become permanent. To keep it, ask the Agent to implement a normal local, project, or repository Plugin through the regular development workflow. It may affect other sessions in the same process; the sandbox is not a security boundary, and injected services reach the real runtime. `code` runs now as the body of an async JavaScript function in an isolated sandbox and MUST `return` a plugin. Two forms: FUNCTION form `return (ctx) => { … }` — declares no inject, so it can register tools, listen to events, and provide services, but reaching ANY service (e.g. ctx.bash) throws; use it only when you need no services. OBJECT form `return { name?, inject: ['bash', 'llm', …], apply(ctx) { … } }` — declares dependencies, and cordis activates the plugin only after the services exist; PREFER this form. You may reach ONLY the services you list in inject: an undeclared service throws even if it exists, because an undeclared dependency would not be cleaned up if its provider is unmounted. BEFORE calling a service from your code, read cordis_inspect what:\"api\" — it lists method signatures AND the type shapes of their arguments/returns (do not guess a field's type; e.g. a bash run's stdout is an object, not a string). Inside `apply`, use the standard cordis API: `ctx.on(event, listener)` to observe events (see cordis_inspect what:\"events\"), or call `harness.registerTool(ctx, harness.defineTool({ name, description, parameters: { text: { type: 'string', required: true } }, output: { schema: { type: 'string' }, render(_args, value) { return [{ type: 'text', text: value }] } }, async execute(args) { return args.text } }))` to give yourself a new tool — it becomes callable on your NEXT step. Tool parameters: each key IS a property — { type: 'string'|'number'|'integer'|'boolean'|'null'|'object'|'array'|'json', required?: true, description?, enum?, const?, items?, properties? }; every direct DSL object declares additionalProperties: true|false, and oneOf: [schema, schema, ...] replaces type for an exact-one union. A raw JSON-Schema { type: 'object', properties, required?: […] } wrapper is also accepted with open-by-default objects. A tool's `execute` MUST return the lossless JSON value declared by `output.schema`; `output.render(args, value)` separately returns Native/model content blocks. Temporary Plugins can COMPOSE: one Plugin may `ctx.provide('name', value)` a service and another may declare `inject: ['name']` to consume it — the consumer stays pending until the provider exists and returns to pending when the provider is unmounted. Everything registered inside `apply` is cleaned up automatically by cordis_unmount. Sandbox globals: `console` (tagged `[cordis:]`, writes through to the harness terminal), `harness.defineTool`, `harness.registerTool`, `btoa`, `atob`, `TextEncoder`, `TextDecoder`. Node APIs are DISABLED — do filesystem/network/timer work through the cordis services, never Node built-ins: `require`, `setTimeout`/`setInterval`, and `fetch` throw redirect errors; `process` and `Buffer` are undefined. Instead use inject: ['fs'] + ctx.fs for files, inject: ['web'] + ctx.web for HTTP, inject: ['bash'] + ctx.bash for processes, and inject: ['timer'] + ctx.setTimeout/ctx.setInterval for timing (fiber effects, auto-cleaned when unmounted) — cordis_inspect what:\"api\" shows what THIS runtime provides. Write PLAIN JavaScript, not TypeScript (no `as`, no type annotations). Cautions: (1) waterfall events (e.g. tools/pre-execute) hand the listener a trailing `next` callback which MUST be called — returning without `next()` VETOES the call; prefer plain notification events unless you intend to intercept. (2) Never await something that only resolves after the current turn (your code runs INSIDE a tool call of that turn — it would deadlock). (3) Your `ctx` is a restricted façade: you can register tools, observe events, provide/consume services, and use timers, but framework internals (ctx.root, ctx.fiber, ctx.extend, ctx.plugin, …) are withheld. It is not a security boundary though — the services you inject (e.g. ctx.bash) reach the real runtime. */\n cordis_mount: {\n /** JavaScript body returning a temporary Plugin; evaluated now and saved nowhere. */\n code: string;\n } & Record;\n /** Unmount a current-process temporary Plugin created by cordis_mount. Waits for its tools, listeners, services, timers, and other owned effects to clean up completely. Only dyn-N temporary ids are accepted; this cannot remove Loader, configured, or installed Plugins. */\n cordis_unmount: {\n /** The temporary Plugin id returned by cordis_mount (for example \"dyn-1\"); valid only in this process and invalid after unmount or restart. */\n id: string;\n } & Record;\n /** Edit an existing UTF-8 text file by replacing literal text. */\n edit: {\n /** Path to edit, resolved by the filesystem backend. */\n file_path: string;\n /** Literal text to replace. Must match exactly. */\n old_string: string;\n /** Literal replacement text. Use an empty string to delete the match. */\n new_string: string;\n /** Replace all matches. Defaults to false; when false, old_string must appear exactly once. */\n replace_all?: boolean;\n } & Record;\n /** Run a foreground fresh-agent Ralph loop toward one immutable objective. Use only when the direct human explicitly asks for Ralph or fresh-agent iteration. Each round opens a new child with no parent conversation or prior child session; the shared workspace is long-term memory, and only a bounded structured report crosses rounds. The call returns when a worker reports completion or a concrete blocker, or at the round limit. Ordinary long-running same-session work belongs to goal tools. */\n ralph: {\n /** The immutable completion objective for every fresh Ralph round. */\n objective: string;\n /** Optional positive safe-integer round cap, bounded by the deployment ceiling. */\n maxRounds?: number;\n } & Record;\n /** Read a UTF-8 text file and return line-numbered content. */\n read: {\n /** Path to read, resolved by the filesystem backend. */\n file_path: string;\n /** 1-based first line to return. Defaults to 1. */\n offset?: number;\n /** Maximum number of lines to return. Defaults to 2000. */\n limit?: number;\n } & Record;\n /** Load the full instructions for an available skill. Call this with the exact skill name from the session skill catalog before acting on a task that names or clearly matches that skill. */\n skill: {\n /** The exact skill name from the available skills list. */\n name: string;\n } & Record;\n /** Delegate a self-contained task to a subagent (a separate agent that works in its own context) and return its final result. Use this to offload focused, independent work — research, a scoped implementation, an analysis — so it does not consume this conversation's context. The subagent runs to completion and you receive only its final answer, not its intermediate steps. Give it a complete, standalone prompt: it does not see this conversation. Set `run_in_background: true` to return a task id; collect with `task_output` and stop with `task_kill`. */\n subagent: {\n /** A short (3-5 word) description of the delegated task, for display. */\n description: string;\n /** The complete, self-contained task for the subagent. It does not share this conversation's context, so include everything it needs. */\n prompt: string;\n /** Run as a background task and return its id; collect with task_output or stop with task_kill. */\n run_in_background?: boolean;\n } & Record;\n /** Delegate a task to a subagent that inherits this conversation: a child agent seeded with all completed turns so far (it does not see the current in-flight turn), returning only its final result. Use this when the subtask builds on this conversation's context — a follow-up analysis, a review, a continuation — without consuming this conversation's context for the work itself. You receive only its final answer, not its intermediate steps. Set `run_in_background: true` to return a task id; collect with `task_output` and stop with `task_kill`. */\n subagent_fork: {\n /** A short (3-5 word) description of the delegated task, for display. */\n description: string;\n /** The task for the subagent. It already sees this conversation's completed turns, so build on them freely and state only what is new. */\n prompt: string;\n /** Run as a background task and return its id; collect with task_output or stop with task_kill. */\n run_in_background?: boolean;\n } & Record;\n /** Request cancellation of a running background task by task id. Returns immediately; the task settles as killed once its work actually stops. */\n task_kill: {\n /** Task id returned by the tool that started the background work. */\n task_id: string;\n /** Optional short reason, recorded in the log and forwarded to the task. */\n reason?: string;\n } & Record;\n /** List your background tasks (running and finished) with their ids, kinds, and statuses. */\n task_list: Record;\n /** Read a background task. Stream tasks return only output since the previous read; final-output tasks return their result after settlement. Every response ends with `[status: ...]`. Reads are non-blocking unless `wait: true`, which waits up to the configured cap. */\n task_output: {\n /** Task id returned by the tool that started the background work. */\n task_id: string;\n /** Block until the task reaches a terminal status or the timeout expires. A timed-out wait returns [status: running] and leaves the task alive. */\n wait?: boolean;\n /** Max wait in milliseconds (only meaningful with wait: true). Defaults to the configured wait timeout; capped by the configured maximum. */\n timeout_ms?: number;\n } & Record;\n /** Record and update a structured task list for the current work. Send the ENTIRE list every call — it REPLACES the previous list (there are no partial updates, no per-item edits). Use it to plan multi-step work and show progress: add one todo per concrete step before you start. Keep AT MOST ONE todo `in_progress` at a time; while work remains, exactly one active task should be `in_progress`. Mark a todo `completed` the moment it is done (do not batch completions), and allow no `in_progress` item only once all work is complete. Skip the list for trivial single-step tasks. Statuses: `pending` (not started), `in_progress` (being worked on now), `completed` (finished). */\n todo_write: {\n /** The COMPLETE task list, replacing any previous list. */\n todos: ({\n /** What the task is — a short imperative line. */\n content: string;\n /** pending (not started) | in_progress (now) | completed (done). */\n status: \"pending\" | \"in_progress\" | \"completed\";\n })[];\n } & Record;\n /** Run a JavaScript workflow script that orchestrates subagents at scale. Use this for work that fans out across many independent pieces — an audit over many files, a migration, multi-angle research, adversarial verification of findings — where you write the orchestration as a script instead of delegating turn by turn. The workflow's identity rides the `meta` parameter as JSON: required `name` (short kebab-case) and `description` strings, optional `whenToUse` string and `phases` array (`{title, detail?, provider?, model?}`). The `script` parameter is the plain JavaScript body ONLY (NOT TypeScript, and NO `export const meta` statement — meta is a parameter, not code), running with top-level await; end with `return ` — the value must be JSON-serializable and is this tool's result. Script-body hooks: - `agent(prompt, opts?): Promise` — run one subagent to completion. Without `opts.schema` it resolves to the child's final text; with `opts.schema` (an object-rooted JSON Schema using ONLY type/properties/required/additionalProperties/items/enum/const/oneOf — no pattern/format/numeric bounds) it resolves to the validated object. Resolves `null` when the child fails (filter with `.filter(Boolean)`). Other opts: `label` (display), `phase` (progress group), and independent `provider`/`model` LLM target overrides (either may be provided alone). Anything else (`effort`/`isolation`/`agentType`) is rejected loudly. - `pipeline(items, ...stages): Promise` — run each item through the stages independently with NO barrier between stages (prefer this for multi-stage work). Each stage receives `(prev, item, index)`. An ordinary stage throw drops that ITEM to `null` and skips its remaining stages. - `parallel(thunks): Promise` — run zero-argument functions concurrently and await ALL of them (a barrier; use only when a stage genuinely needs every prior result together). A throwing thunk resolves to `null`. - `phase(title)` — start a progress phase; `log(message)` — narrate progress; `args` — the tool call's `args` input, verbatim. Misused hooks (bad arguments, unknown options, unsupported schemas, tripped caps) throw errors that ALWAYS kill the script — they never dissolve into a per-item `null`. Constraints: concurrency and total-agent caps apply; no filesystem, network, timers, or Node.js APIs are provided — the agents do the work, the script only coordinates them. The run executes in the foreground: this call returns when the whole script finishes. */\n workflow: {\n /** The plain-JS workflow script body (top-level await allowed; NO `export const meta` statement; end with `return `). */\n script: string;\n /** The workflow identity block (plain JSON — never code). */\n meta: {\n /** Short kebab-case workflow name. */\n name: string;\n /** One-line description of what the workflow does. */\n description: string;\n /** Optional guidance on when this workflow applies. */\n whenToUse?: string;\n /** Optional phase declarations matched by phase() calls. */\n phases?: ({\n /** The phase title phase() calls match by exact string. */\n title: string;\n /** Optional one-line description of the phase. */\n detail?: string;\n /** Optional provider override this phase is expected to use. */\n provider?: string;\n /** Optional model override this phase is expected to use. */\n model?: string;\n } & Record)[];\n } & Record;\n /** Optional JSON input exposed to the script as the `args` global (wrap a bare list as a field, e.g. {\"files\": [...]}). */\n args?: Record;\n } & Record;\n /** Create or fully replace a UTF-8 text file. */\n write: {\n /** Path to write, resolved by the filesystem backend. */\n file_path: string;\n /** Full UTF-8 text content to write. */\n content: string;\n } & Record;\n}\n\ninterface ToolOutputMap {\n bash: {\n kind: \"background\";\n taskId: string;\n } | {\n kind: \"foreground\";\n exitCode: number | null;\n signal: string | null;\n timedOut: boolean;\n aborted: boolean;\n timeoutMs: number;\n stdout: {\n text: string;\n truncated: boolean;\n spillPath?: string;\n };\n stderr: {\n text: string;\n truncated: boolean;\n spillPath?: string;\n };\n sandbox?: {\n mode: string;\n denied: boolean;\n enforcement?: string;\n runnerFailed?: boolean;\n };\n };\n cordis_inspect: string;\n cordis_mount: {\n id: string;\n pluginName: string;\n state: \"pending\" | \"loading\" | \"active\" | \"failed\" | \"disposed\" | \"unloading\";\n provides: string[];\n waitingFor: string[];\n };\n cordis_unmount: {\n id: string;\n pluginName: string;\n };\n edit: {\n path: string;\n before: string;\n after: string;\n };\n ralph: {\n runId: string;\n agentsStarted: number;\n result: JsonValue;\n };\n read: {\n path: string;\n offset: number;\n lines: {\n number: number;\n text: string;\n }[];\n totalLines: number;\n };\n skill: {\n name: string;\n provider: string;\n resourceBase?: {\n kind: \"directory\";\n path: string;\n } | {\n kind: \"url\";\n url: string;\n } | {\n kind: \"opaque\";\n description: string;\n };\n content: string;\n };\n subagent: {\n kind: \"background\";\n taskId: string;\n } | {\n kind: \"foreground\";\n runId: string;\n output: JsonValue[];\n };\n subagent_fork: {\n kind: \"background\";\n taskId: string;\n } | {\n kind: \"foreground\";\n runId: string;\n output: JsonValue[];\n };\n task_kill: {\n outcome: \"cancellation-requested\" | \"already-finished\";\n task: {\n id: string;\n kind: string;\n label: string;\n status: \"running\" | \"stopping\" | \"completed\" | \"killed\" | \"failed\";\n detail?: string;\n startedAt: number;\n finishedAt?: number;\n };\n };\n task_list: ({\n id: string;\n kind: string;\n label: string;\n status: \"running\" | \"stopping\" | \"completed\" | \"killed\" | \"failed\";\n detail?: string;\n startedAt: number;\n finishedAt?: number;\n })[];\n task_output: {\n text: string;\n task: {\n id: string;\n kind: string;\n label: string;\n status: \"running\" | \"stopping\" | \"completed\" | \"killed\" | \"failed\";\n detail?: string;\n startedAt: number;\n finishedAt?: number;\n };\n };\n todo_write: {\n todos: ({\n content: string;\n status: \"pending\" | \"in_progress\" | \"completed\";\n })[];\n counts: {\n pending: number;\n inProgress: number;\n completed: number;\n };\n };\n workflow: {\n runId: string;\n agentsStarted: number;\n result: JsonValue;\n };\n write: {\n path: string;\n operation: \"create\" | \"update\";\n before: string | null;\n after: string;\n };\n}\n\ntype ToolName = keyof ToolOutputMap\n\ndeclare class ToolCallError extends Error {\n readonly name: \"ToolCallError\";\n readonly toolName: ToolName;\n}\n\ndeclare const tools: {\n [K in ToolName]: (args: ToolArgsMap[K]) => Promise;\n}\n```","tools":[{"name":"bash","description":"Execute a bash command (`bash -c`) and return its stdout/stderr. Each call runs in a fresh shell: no state (cwd, variables, functions) persists between calls — pass `workdir` instead of using `cd`. Non-zero exits are reported as `[exit code: N]`. Current harness environment facts are exposed through managed `$DSH_*` variables; inspect them when needed. Commands may run under a file sandbox; a blocked file operation is reported as `[sandbox: file access denied under mode]` — a policy denial, not a bug in the command; do not retry another way. Long output is truncated to its tail; the full output is saved to a file whose path is reported when available. Set `run_in_background: true` for long-running commands: the call returns a task id immediately; read its output with `task_output` and stop it with `task_kill`.","parameters":{"type":"object","properties":{"command":{"type":"string","description":"The bash command to execute."},"description":{"type":"string","description":"Clear, concise description of what this command does in active voice, 5-10 words (shown in the UI). Examples: \"ls\" → \"List files in current directory\"; \"git status\" → \"Show working tree status\"; \"npm install\" → \"Install package dependencies\"."},"timeoutMs":{"type":"number","description":"Timeout in milliseconds. The executor applies its configured default and cap, and kills the command on expiry."},"workdir":{"type":"string","description":"Working directory for this command. Defaults to the session workspace; a relative path is resolved against it."},"run_in_background":{"type":"boolean","description":"Run in the background and return a task id immediately (collect with task_output, stop with task_kill). No timeout applies."}},"required":["command","description"]}},{"name":"cordis_inspect","description":"Inspect the live Cordis runtime in the current DSH process. Read-only. Sections: `services` (every provided ctx service and the plugin fiber that owns it), `plugins` (all live plugin fibers with their lifecycle states), `tools` (the model-facing tools currently registered, i.e. what you can call), `temporary` (only temporary Plugins created by cordis_mount: id, name, state, provided services, awaited services, and lifetime), `api` (method signatures AND argument/return type shapes for every LIVE service — read this before writing plugin code that calls a service), `events` (every harness event with its dispatch mode and exact signature — pick listener targets here). Temporary Plugins exist only in memory, remain active across later turns, and disappear after cordis_unmount, toolset unload, or DSH restart; they are not restored automatically. The `temporary` section is a subset of `plugins`. Omit `what` to get all six sections. With `what:\"api\"` or `what:\"events\"`, pass an exact `name` to narrow to one service/event and include its original source JSDoc.","parameters":{"type":"object","properties":{"what":{"type":"string","description":"Limit the report to one section. Omit for all sections.","enum":["services","plugins","tools","temporary","api","events"]},"name":{"type":"string","description":"Exact service key or event name whose original JSDoc to include; valid only with what:\"api\" or what:\"events\"."}}}},{"name":"cordis_mount","description":"Mount a temporary Cordis Plugin in the current DSH process. This creates an in-memory runtime Plugin, not an installed or configured Plugin. It remains active across later turns until cordis_unmount, toolset unload, or DSH restart. It does not create files, install a package, change cordis.yml or personal/project config, survive restart, or automatically become permanent. To keep it, ask the Agent to implement a normal local, project, or repository Plugin through the regular development workflow. It may affect other sessions in the same process; the sandbox is not a security boundary, and injected services reach the real runtime. `code` runs now as the body of an async JavaScript function in an isolated sandbox and MUST `return` a plugin. Two forms: FUNCTION form `return (ctx) => { … }` — declares no inject, so it can register tools, listen to events, and provide services, but reaching ANY service (e.g. ctx.bash) throws; use it only when you need no services. OBJECT form `return { name?, inject: ['bash', 'llm', …], apply(ctx) { … } }` — declares dependencies, and cordis activates the plugin only after the services exist; PREFER this form. You may reach ONLY the services you list in inject: an undeclared service throws even if it exists, because an undeclared dependency would not be cleaned up if its provider is unmounted. BEFORE calling a service from your code, read cordis_inspect what:\"api\" — it lists method signatures AND the type shapes of their arguments/returns (do not guess a field's type; e.g. a bash run's stdout is an object, not a string). Inside `apply`, use the standard cordis API: `ctx.on(event, listener)` to observe events (see cordis_inspect what:\"events\"), or call `harness.registerTool(ctx, harness.defineTool({ name, description, parameters: { text: { type: 'string', required: true } }, output: { schema: { type: 'string' }, render(_args, value) { return [{ type: 'text', text: value }] } }, async execute(args) { return args.text } }))` to give yourself a new tool — it becomes callable on your NEXT step. Tool parameters: each key IS a property — { type: 'string'|'number'|'integer'|'boolean'|'null'|'object'|'array'|'json', required?: true, description?, enum?, const?, items?, properties? }; every direct DSL object declares additionalProperties: true|false, and oneOf: [schema, schema, ...] replaces type for an exact-one union. A raw JSON-Schema { type: 'object', properties, required?: […] } wrapper is also accepted with open-by-default objects. A tool's `execute` MUST return the lossless JSON value declared by `output.schema`; `output.render(args, value)` separately returns Native/model content blocks. Temporary Plugins can COMPOSE: one Plugin may `ctx.provide('name', value)` a service and another may declare `inject: ['name']` to consume it — the consumer stays pending until the provider exists and returns to pending when the provider is unmounted. Everything registered inside `apply` is cleaned up automatically by cordis_unmount. Sandbox globals: `console` (tagged `[cordis:]`, writes through to the harness terminal), `harness.defineTool`, `harness.registerTool`, `btoa`, `atob`, `TextEncoder`, `TextDecoder`. Node APIs are DISABLED — do filesystem/network/timer work through the cordis services, never Node built-ins: `require`, `setTimeout`/`setInterval`, and `fetch` throw redirect errors; `process` and `Buffer` are undefined. Instead use inject: ['fs'] + ctx.fs for files, inject: ['web'] + ctx.web for HTTP, inject: ['bash'] + ctx.bash for processes, and inject: ['timer'] + ctx.setTimeout/ctx.setInterval for timing (fiber effects, auto-cleaned when unmounted) — cordis_inspect what:\"api\" shows what THIS runtime provides. Write PLAIN JavaScript, not TypeScript (no `as`, no type annotations). Cautions: (1) waterfall events (e.g. tools/pre-execute) hand the listener a trailing `next` callback which MUST be called — returning without `next()` VETOES the call; prefer plain notification events unless you intend to intercept. (2) Never await something that only resolves after the current turn (your code runs INSIDE a tool call of that turn — it would deadlock). (3) Your `ctx` is a restricted façade: you can register tools, observe events, provide/consume services, and use timers, but framework internals (ctx.root, ctx.fiber, ctx.extend, ctx.plugin, …) are withheld. It is not a security boundary though — the services you inject (e.g. ctx.bash) reach the real runtime.","parameters":{"type":"object","properties":{"code":{"type":"string","description":"JavaScript body returning a temporary Plugin; evaluated now and saved nowhere."}},"required":["code"]}},{"name":"cordis_unmount","description":"Unmount a current-process temporary Plugin created by cordis_mount. Waits for its tools, listeners, services, timers, and other owned effects to clean up completely. Only dyn-N temporary ids are accepted; this cannot remove Loader, configured, or installed Plugins.","parameters":{"type":"object","properties":{"id":{"type":"string","description":"The temporary Plugin id returned by cordis_mount (for example \"dyn-1\"); valid only in this process and invalid after unmount or restart."}},"required":["id"]}},{"name":"edit","description":"Edit an existing UTF-8 text file by replacing literal text.","parameters":{"type":"object","properties":{"file_path":{"type":"string","description":"Path to edit, resolved by the filesystem backend."},"old_string":{"type":"string","description":"Literal text to replace. Must match exactly."},"new_string":{"type":"string","description":"Literal replacement text. Use an empty string to delete the match."},"replace_all":{"type":"boolean","description":"Replace all matches. Defaults to false; when false, old_string must appear exactly once."}},"required":["file_path","old_string","new_string"]}},{"name":"ralph","description":"Run a foreground fresh-agent Ralph loop toward one immutable objective. Use only when the direct human explicitly asks for Ralph or fresh-agent iteration. Each round opens a new child with no parent conversation or prior child session; the shared workspace is long-term memory, and only a bounded structured report crosses rounds. The call returns when a worker reports completion or a concrete blocker, or at the round limit. Ordinary long-running same-session work belongs to goal tools.","parameters":{"type":"object","properties":{"objective":{"type":"string","description":"The immutable completion objective for every fresh Ralph round."},"maxRounds":{"type":"number","description":"Optional positive safe-integer round cap, bounded by the deployment ceiling."}},"required":["objective"]}},{"name":"read","description":"Read a UTF-8 text file and return line-numbered content.","parameters":{"type":"object","properties":{"file_path":{"type":"string","description":"Path to read, resolved by the filesystem backend."},"offset":{"type":"number","description":"1-based first line to return. Defaults to 1."},"limit":{"type":"number","description":"Maximum number of lines to return. Defaults to 2000."}},"required":["file_path"]}},{"name":"run_code","description":"Execute a TypeScript program against the available tools. Write the BODY of an async function (erasable syntax only; top-level `await` and `return` work) and call tools as `await tools.name(args)` per the declarations in the system prompt. Only what you print or return comes back — curate it.","parameters":{"type":"object","properties":{"code":{"type":"string","description":"The program: the body of an async TypeScript function."},"description":{"type":"string","description":"Clear, concise description of what this program does in active voice, 5-10 words (shown in the UI). Examples: \"Count TODO markers across packages\"; \"Read failing test and its fixture\"; \"Rename config key in every cordis.yml\"."}},"required":["code","description"]}},{"name":"skill","description":"Load the full instructions for an available skill. Call this with the exact skill name from the session skill catalog before acting on a task that names or clearly matches that skill.","parameters":{"type":"object","properties":{"name":{"type":"string","description":"The exact skill name from the available skills list."}},"required":["name"]}},{"name":"subagent","description":"Delegate a self-contained task to a subagent (a separate agent that works in its own context) and return its final result. Use this to offload focused, independent work — research, a scoped implementation, an analysis — so it does not consume this conversation's context. The subagent runs to completion and you receive only its final answer, not its intermediate steps. Give it a complete, standalone prompt: it does not see this conversation. Set `run_in_background: true` to return a task id; collect with `task_output` and stop with `task_kill`.","parameters":{"type":"object","properties":{"description":{"type":"string","description":"A short (3-5 word) description of the delegated task, for display."},"prompt":{"type":"string","description":"The complete, self-contained task for the subagent. It does not share this conversation's context, so include everything it needs."},"run_in_background":{"type":"boolean","description":"Run as a background task and return its id; collect with task_output or stop with task_kill."}},"required":["description","prompt"]}},{"name":"subagent_fork","description":"Delegate a task to a subagent that inherits this conversation: a child agent seeded with all completed turns so far (it does not see the current in-flight turn), returning only its final result. Use this when the subtask builds on this conversation's context — a follow-up analysis, a review, a continuation — without consuming this conversation's context for the work itself. You receive only its final answer, not its intermediate steps. Set `run_in_background: true` to return a task id; collect with `task_output` and stop with `task_kill`.","parameters":{"type":"object","properties":{"description":{"type":"string","description":"A short (3-5 word) description of the delegated task, for display."},"prompt":{"type":"string","description":"The task for the subagent. It already sees this conversation's completed turns, so build on them freely and state only what is new."},"run_in_background":{"type":"boolean","description":"Run as a background task and return its id; collect with task_output or stop with task_kill."}},"required":["description","prompt"]}},{"name":"task_kill","description":"Request cancellation of a running background task by task id. Returns immediately; the task settles as killed once its work actually stops.","parameters":{"type":"object","properties":{"task_id":{"type":"string","description":"Task id returned by the tool that started the background work."},"reason":{"type":"string","description":"Optional short reason, recorded in the log and forwarded to the task."}},"required":["task_id"]}},{"name":"task_list","description":"List your background tasks (running and finished) with their ids, kinds, and statuses.","parameters":{"type":"object","properties":{}}},{"name":"task_output","description":"Read a background task. Stream tasks return only output since the previous read; final-output tasks return their result after settlement. Every response ends with `[status: ...]`. Reads are non-blocking unless `wait: true`, which waits up to the configured cap.","parameters":{"type":"object","properties":{"task_id":{"type":"string","description":"Task id returned by the tool that started the background work."},"wait":{"type":"boolean","description":"Block until the task reaches a terminal status or the timeout expires. A timed-out wait returns [status: running] and leaves the task alive."},"timeout_ms":{"type":"number","description":"Max wait in milliseconds (only meaningful with wait: true). Defaults to the configured wait timeout; capped by the configured maximum."}},"required":["task_id"]}},{"name":"todo_write","description":"Record and update a structured task list for the current work. Send the ENTIRE list every call — it REPLACES the previous list (there are no partial updates, no per-item edits). Use it to plan multi-step work and show progress: add one todo per concrete step before you start. Keep AT MOST ONE todo `in_progress` at a time; while work remains, exactly one active task should be `in_progress`. Mark a todo `completed` the moment it is done (do not batch completions), and allow no `in_progress` item only once all work is complete. Skip the list for trivial single-step tasks. Statuses: `pending` (not started), `in_progress` (being worked on now), `completed` (finished).","parameters":{"type":"object","properties":{"todos":{"type":"array","description":"The COMPLETE task list, replacing any previous list.","items":{"type":"object","additionalProperties":false,"properties":{"content":{"type":"string","description":"What the task is — a short imperative line."},"status":{"type":"string","description":"pending (not started) | in_progress (now) | completed (done).","enum":["pending","in_progress","completed"]}},"required":["content","status"]}}},"required":["todos"]}},{"name":"workflow","description":"Run a JavaScript workflow script that orchestrates subagents at scale. Use this for work that fans out across many independent pieces — an audit over many files, a migration, multi-angle research, adversarial verification of findings — where you write the orchestration as a script instead of delegating turn by turn.\n\nThe workflow's identity rides the `meta` parameter as JSON: required `name` (short kebab-case) and `description` strings, optional `whenToUse` string and `phases` array (`{title, detail?, provider?, model?}`). The `script` parameter is the plain JavaScript body ONLY (NOT TypeScript, and NO `export const meta` statement — meta is a parameter, not code), running with top-level await; end with `return ` — the value must be JSON-serializable and is this tool's result.\n\nScript-body hooks:\n- `agent(prompt, opts?): Promise` — run one subagent to completion. Without `opts.schema` it resolves to the child's final text; with `opts.schema` (an object-rooted JSON Schema using ONLY type/properties/required/additionalProperties/items/enum/const/oneOf — no pattern/format/numeric bounds) it resolves to the validated object. Resolves `null` when the child fails (filter with `.filter(Boolean)`). Other opts: `label` (display), `phase` (progress group), and independent `provider`/`model` LLM target overrides (either may be provided alone). Anything else (`effort`/`isolation`/`agentType`) is rejected loudly.\n- `pipeline(items, ...stages): Promise` — run each item through the stages independently with NO barrier between stages (prefer this for multi-stage work). Each stage receives `(prev, item, index)`. An ordinary stage throw drops that ITEM to `null` and skips its remaining stages.\n- `parallel(thunks): Promise` — run zero-argument functions concurrently and await ALL of them (a barrier; use only when a stage genuinely needs every prior result together). A throwing thunk resolves to `null`.\n- `phase(title)` — start a progress phase; `log(message)` — narrate progress; `args` — the tool call's `args` input, verbatim.\n\nMisused hooks (bad arguments, unknown options, unsupported schemas, tripped caps) throw errors that ALWAYS kill the script — they never dissolve into a per-item `null`.\n\nConstraints: concurrency and total-agent caps apply; no filesystem, network, timers, or Node.js APIs are provided — the agents do the work, the script only coordinates them. The run executes in the foreground: this call returns when the whole script finishes.","parameters":{"type":"object","properties":{"script":{"type":"string","description":"The plain-JS workflow script body (top-level await allowed; NO `export const meta` statement; end with `return `)."},"meta":{"type":"object","description":"The workflow identity block (plain JSON — never code).","additionalProperties":true,"properties":{"name":{"type":"string","description":"Short kebab-case workflow name."},"description":{"type":"string","description":"One-line description of what the workflow does."},"whenToUse":{"type":"string","description":"Optional guidance on when this workflow applies."},"phases":{"type":"array","description":"Optional phase declarations matched by phase() calls.","items":{"type":"object","additionalProperties":true,"properties":{"title":{"type":"string","description":"The phase title phase() calls match by exact string."},"detail":{"type":"string","description":"Optional one-line description of the phase."},"provider":{"type":"string","description":"Optional provider override this phase is expected to use."},"model":{"type":"string","description":"Optional model override this phase is expected to use."}},"required":["title"]}}},"required":["name","description"]},"args":{"type":"object","description":"Optional JSON input exposed to the script as the `args` global (wrap a bare list as a field, e.g. {\"files\": [...]}).","additionalProperties":true}},"required":["script","meta"]}},{"name":"write","description":"Create or fully replace a UTF-8 text file.","parameters":{"type":"object","properties":{"file_path":{"type":"string","description":"Path to write, resolved by the filesystem backend."},"content":{"type":"string","description":"Full UTF-8 text content to write."}},"required":["file_path","content"]}}]},"reason":"initial"}} +{"type":"request/header","seq":4,"time":1783957884486,"data":{"header":{"config":{"provider":"deepseek","model":"deepseek-v4-flash"},"system":"You are an AI agent powered by the DeepSeek Harness SDK.\n\nYou are headless-agent, a coding assistant powered by the deepseek-v4-flash model. Your working directory is {{cwd}}.\n\nVerify your work by running the code or tests. Keep answers brief and factual.\n\n\nUse the read tool — not shell commands like cat — to inspect text files. Results include line numbers. Use offset and limit to continue reading large files.\n\nUse the write tool to create files or completely replace file contents. Existing files are overwritten, so read an existing file first (the default fs-policy requires it) and prefer edit for targeted changes.\n\nUse the edit tool for targeted changes to existing UTF-8 text files. It replaces literal old_string with new_string; by default old_string must appear exactly once. If old_string appears multiple times, provide a more specific old_string or set replace_all to true. Read the file first (the default fs-policy requires it), unless you just created or edited it in this session.\n\nCheck the [exit code: N] marker on every bash result; investigate failures before moving on.\n\nTrack every background task id you start. You are notified in-session when a task finishes — do not busy-poll or sleep on one; keep working on independent steps and do not duplicate a running task's work. Before giving a final answer, collect every still-relevant task with task_output (set wait: true only when you are genuinely blocked on it), and task_kill tasks that stopped mattering.\n\nUse the workflow tool ONLY when the user explicitly asks for a workflow or for large multi-agent orchestration: you write a JavaScript script (the tool description documents the exact format) that fans work out across many subagents with phases and structured results. For one or two delegations, prefer plain subagent calls.\n\nUse the ralph tool ONLY when the direct human explicitly asks for a Ralph loop or fresh-agent iterative execution. Each Ralph round starts a fresh child with no conversation seed and uses the shared workspace as durable memory. Completion and blockers are worker reports, not independent evaluation. Use same-session goal tools for ordinary long-running objectives, and plain subagents or workflows for bounded delegation and fan-out.\n\n## Writing code for run_code\n\nPass `run_code` the body of an async TypeScript function (erasable syntax only — no `enum` or namespaces; type annotations are advisory, the code runs type-stripped). Inside the program:\n\n- Call tools as `await tools.name(args)` — quoted access for exotic names: `tools[\"my-tool\"](args)`. Every call resolves to the tool's typed canonical JSON value. Tool arguments must be lossless JSON.\n- A FAILED tool call rejects with `ToolCallError`, whose `toolName` identifies the failed tool and whose `message` is human-readable — `try/catch` it to handle and continue.\n- Independent read-only calls MAY overlap under `Promise.all` (safe calls run concurrently; mutating calls run alone, in submission order). Sequence dependent work with `await`.\n- Emit results with `return` and/or `console.log(...)`. ONLY what you print or return comes back to you — intermediate tool results never enter the conversation, so extract just what you need.\n\nThe available tools:\n\n```ts\ntype JsonValue = null | boolean | number | string | JsonValue[] | { [key: string]: JsonValue }\n\ninterface ToolArgsMap {\n /** Execute a bash command (`bash -c`) and return its stdout/stderr. Each call runs in a fresh shell: no state (cwd, variables, functions) persists between calls — pass `workdir` instead of using `cd`. Non-zero exits are reported as `[exit code: N]`. Current harness environment facts are exposed through managed `$DSH_*` variables; inspect them when needed. Commands may run under a file sandbox; a blocked file operation is reported as `[sandbox: file access denied under mode]` — a policy denial, not a bug in the command; do not retry another way. Long output is truncated to its tail; the full output is saved to a file whose path is reported when available. Set `run_in_background: true` for long-running commands: the call returns a task id immediately; read its output with `task_output` and stop it with `task_kill`. */\n bash: {\n /** The bash command to execute. */\n command: string;\n /** Clear, concise description of what this command does in active voice, 5-10 words (shown in the UI). Examples: \"ls\" → \"List files in current directory\"; \"git status\" → \"Show working tree status\"; \"npm install\" → \"Install package dependencies\". */\n description: string;\n /** Timeout in milliseconds. The executor applies its configured default and cap, and kills the command on expiry. */\n timeoutMs?: number;\n /** Working directory for this command. Defaults to the session workspace; a relative path is resolved against it. */\n workdir?: string;\n /** Run in the background and return a task id immediately (collect with task_output, stop with task_kill). No timeout applies. */\n run_in_background?: boolean;\n } & Record;\n /** Inspect the live Cordis runtime in the current DSH process. Read-only. Sections: `services` (every provided ctx service and the plugin fiber that owns it), `plugins` (all live plugin fibers with their lifecycle states), `tools` (the model-facing tools currently registered, i.e. what you can call), `temporary` (only temporary Plugins created by cordis_mount: id, name, state, provided services, awaited services, and lifetime), `api` (method signatures AND argument/return type shapes for every LIVE service — read this before writing plugin code that calls a service), `events` (every harness event with its dispatch mode and exact signature — pick listener targets here). Temporary Plugins exist only in memory, remain active across later turns, and disappear after cordis_unmount, toolset unload, or DSH restart; they are not restored automatically. The `temporary` section is a subset of `plugins`. Omit `what` to get all six sections. With `what:\"api\"` or `what:\"events\"`, pass an exact `name` to narrow to one service/event and include its original source JSDoc. */\n cordis_inspect: {\n /** Limit the report to one section. Omit for all sections. */\n what?: \"services\" | \"plugins\" | \"tools\" | \"temporary\" | \"api\" | \"events\";\n /** Exact service key or event name whose original JSDoc to include; valid only with what:\"api\" or what:\"events\". */\n name?: string;\n } & Record;\n /** Mount a temporary Cordis Plugin in the current DSH process. This creates an in-memory runtime Plugin, not an installed or configured Plugin. It remains active across later turns until cordis_unmount, toolset unload, or DSH restart. It does not create files, install a package, change cordis.yml or personal/project config, survive restart, or automatically become permanent. To keep it, ask the Agent to implement a normal local, project, or repository Plugin through the regular development workflow. It may affect other sessions in the same process; the sandbox is not a security boundary, and injected services reach the real runtime. `code` runs now as the body of an async JavaScript function in an isolated sandbox and MUST `return` a plugin. Two forms: FUNCTION form `return (ctx) => { … }` — declares no inject, so it can register tools, listen to events, and provide services, but reaching ANY service (e.g. ctx.bash) throws; use it only when you need no services. OBJECT form `return { name?, inject: ['bash', 'llm', …], apply(ctx) { … } }` — declares dependencies, and cordis activates the plugin only after the services exist; PREFER this form. You may reach ONLY the services you list in inject: an undeclared service throws even if it exists, because an undeclared dependency would not be cleaned up if its provider is unmounted. BEFORE calling a service from your code, read cordis_inspect what:\"api\" — it lists method signatures AND the type shapes of their arguments/returns (do not guess a field's type; e.g. a bash run's stdout is an object, not a string). Inside `apply`, use the standard cordis API: `ctx.on(event, listener)` to observe events (see cordis_inspect what:\"events\"), or call `harness.registerTool(ctx, harness.defineTool({ name, description, parameters: { text: { type: 'string', required: true } }, output: { schema: { type: 'string' }, render(_args, value) { return [{ type: 'text', text: value }] } }, async execute(args) { return args.text } }))` to give yourself a new tool — it becomes callable on your NEXT step. Tool parameters: each key IS a property — { type: 'string'|'number'|'integer'|'boolean'|'null'|'object'|'array'|'json', required?: true, description?, enum?, const?, items?, properties? }; every direct DSL object declares additionalProperties: true|false, and oneOf: [schema, schema, ...] replaces type for an exact-one union. A raw JSON-Schema { type: 'object', properties, required?: […] } wrapper is also accepted with open-by-default objects. A tool's `execute` MUST return the lossless JSON value declared by `output.schema`; `output.render(args, value)` separately returns Native/model content blocks. Temporary Plugins can COMPOSE: one Plugin may `ctx.provide('name', value)` a service and another may declare `inject: ['name']` to consume it — the consumer stays pending until the provider exists and returns to pending when the provider is unmounted. Everything registered inside `apply` is cleaned up automatically by cordis_unmount. Sandbox globals: `console` (tagged `[cordis:]`, writes through to the harness terminal), `harness.defineTool`, `harness.registerTool`, `btoa`, `atob`, `TextEncoder`, `TextDecoder`. Node APIs are DISABLED — do filesystem/network/timer work through the cordis services, never Node built-ins: `require`, `setTimeout`/`setInterval`, and `fetch` throw redirect errors; `process` and `Buffer` are undefined. Instead use inject: ['fs'] + ctx.fs for files, inject: ['web'] + ctx.web for HTTP, inject: ['bash'] + ctx.bash for processes, and inject: ['timer'] + ctx.setTimeout/ctx.setInterval for timing (fiber effects, auto-cleaned when unmounted) — cordis_inspect what:\"api\" shows what THIS runtime provides. Write PLAIN JavaScript, not TypeScript (no `as`, no type annotations). Cautions: (1) waterfall events (e.g. tools/pre-execute) hand the listener a trailing `next` callback which MUST be called — returning without `next()` VETOES the call; prefer plain notification events unless you intend to intercept. (2) Never await something that only resolves after the current turn (your code runs INSIDE a tool call of that turn — it would deadlock). (3) Your `ctx` is a restricted façade: you can register tools, observe events, provide/consume services, and use timers, but framework internals (ctx.root, ctx.fiber, ctx.extend, ctx.plugin, …) are withheld. It is not a security boundary though — the services you inject (e.g. ctx.bash) reach the real runtime. */\n cordis_mount: {\n /** JavaScript body returning a temporary Plugin; evaluated now and saved nowhere. */\n code: string;\n } & Record;\n /** Unmount a current-process temporary Plugin created by cordis_mount. Waits for its tools, listeners, services, timers, and other owned effects to clean up completely. Only dyn-N temporary ids are accepted; this cannot remove Loader, configured, or installed Plugins. */\n cordis_unmount: {\n /** The temporary Plugin id returned by cordis_mount (for example \"dyn-1\"); valid only in this process and invalid after unmount or restart. */\n id: string;\n } & Record;\n /** Edit an existing UTF-8 text file by replacing literal text. */\n edit: {\n /** Path to edit, resolved by the filesystem backend. */\n file_path: string;\n /** Literal text to replace. Must match exactly. */\n old_string: string;\n /** Literal replacement text. Use an empty string to delete the match. */\n new_string: string;\n /** Replace all matches. Defaults to false; when false, old_string must appear exactly once. */\n replace_all?: boolean;\n } & Record;\n /** Run a foreground fresh-agent Ralph loop toward one immutable objective. Use only when the direct human explicitly asks for Ralph or fresh-agent iteration. Each round opens a new child with no parent conversation or prior child session; the shared workspace is long-term memory, and only a bounded structured report crosses rounds. The call returns when a worker reports completion or a concrete blocker, or at the round limit. Ordinary long-running same-session work belongs to goal tools. */\n ralph: {\n /** The immutable completion objective for every fresh Ralph round. */\n objective: string;\n /** Optional positive safe-integer round cap, bounded by the deployment ceiling. */\n maxRounds?: number;\n } & Record;\n /** Read a UTF-8 text file and return line-numbered content. */\n read: {\n /** Path to read, resolved by the filesystem backend. */\n file_path: string;\n /** 1-based first line to return. Defaults to 1. */\n offset?: number;\n /** Maximum number of lines to return. Defaults to 2000. */\n limit?: number;\n } & Record;\n /** Load the full instructions for an available skill. Call this with the exact skill name from the session skill catalog before acting on a task that names or clearly matches that skill. */\n skill: {\n /** The exact skill name from the available skills list. */\n name: string;\n } & Record;\n /** Delegate a self-contained task to a subagent (a separate agent that works in its own context) and return its final result. Use this to offload focused, independent work — research, a scoped implementation, an analysis — so it does not consume this conversation's context. The subagent runs to completion and you receive only its final answer, not its intermediate steps. Give it a complete, standalone prompt: it does not see this conversation. Set `run_in_background: true` to return a task id; collect with `task_output` and stop with `task_kill`. */\n subagent: {\n /** A short (3-5 word) description of the delegated task, for display. */\n description: string;\n /** The complete, self-contained task for the subagent. It does not share this conversation's context, so include everything it needs. */\n prompt: string;\n /** Run as a background task and return its id; collect with task_output or stop with task_kill. */\n run_in_background?: boolean;\n } & Record;\n /** Delegate a task to a subagent that inherits this conversation: a child agent seeded with all completed turns so far (it does not see the current in-flight turn), returning only its final result. Use this when the subtask builds on this conversation's context — a follow-up analysis, a review, a continuation — without consuming this conversation's context for the work itself. You receive only its final answer, not its intermediate steps. Set `run_in_background: true` to return a task id; collect with `task_output` and stop with `task_kill`. */\n subagent_fork: {\n /** A short (3-5 word) description of the delegated task, for display. */\n description: string;\n /** The task for the subagent. It already sees this conversation's completed turns, so build on them freely and state only what is new. */\n prompt: string;\n /** Run as a background task and return its id; collect with task_output or stop with task_kill. */\n run_in_background?: boolean;\n } & Record;\n /** Request cancellation of a running background task by task id. Returns immediately; the task settles as killed once its work actually stops. */\n task_kill: {\n /** Task id returned by the tool that started the background work. */\n task_id: string;\n /** Optional short reason, recorded in the log and forwarded to the task. */\n reason?: string;\n } & Record;\n /** List your background tasks (running and finished) with their ids, kinds, and statuses. */\n task_list: Record;\n /** Read a background task. Stream tasks return only output since the previous read; final-output tasks return their result after settlement. Every response ends with `[status: ...]`. Reads are non-blocking unless `wait: true`, which waits up to the configured cap. */\n task_output: {\n /** Task id returned by the tool that started the background work. */\n task_id: string;\n /** Block until the task reaches a terminal status or the timeout expires. A timed-out wait returns [status: running] and leaves the task alive. */\n wait?: boolean;\n /** Max wait in milliseconds (only meaningful with wait: true). Defaults to the configured wait timeout; capped by the configured maximum. */\n timeout_ms?: number;\n } & Record;\n /** Record and update a structured task list for the current work. Send the ENTIRE list every call — it REPLACES the previous list (there are no partial updates, no per-item edits). Use it to plan multi-step work and show progress: add one todo per concrete step before you start. Keep AT MOST ONE todo `in_progress` at a time; while work remains, exactly one active task should be `in_progress`. Mark a todo `completed` the moment it is done (do not batch completions), and allow no `in_progress` item only once all work is complete. Skip the list for trivial single-step tasks. Statuses: `pending` (not started), `in_progress` (being worked on now), `completed` (finished). */\n todo_write: {\n /** The COMPLETE task list, replacing any previous list. */\n todos: ({\n /** What the task is — a short imperative line. */\n content: string;\n /** pending (not started) | in_progress (now) | completed (done). */\n status: \"pending\" | \"in_progress\" | \"completed\";\n })[];\n } & Record;\n /** Run a JavaScript workflow script that orchestrates subagents at scale. Use this for work that fans out across many independent pieces — an audit over many files, a migration, multi-angle research, adversarial verification of findings — where you write the orchestration as a script instead of delegating turn by turn. The workflow's identity rides the `meta` parameter as JSON: required `name` (short kebab-case) and `description` strings, optional `whenToUse` string and `phases` array (`{title, detail?, provider?, model?}`). The `script` parameter is the plain JavaScript body ONLY (NOT TypeScript, and NO `export const meta` statement — meta is a parameter, not code), running with top-level await; end with `return ` — the value must be JSON-serializable and is this tool's result. Script-body hooks: - `agent(prompt, opts?): Promise` — run one subagent to completion. Without `opts.schema` it resolves to the child's final text; with `opts.schema` (an object-rooted JSON Schema using ONLY type/properties/required/additionalProperties/items/enum/const/oneOf — no pattern/format/numeric bounds) it resolves to the validated object. Resolves `null` when the child fails (filter with `.filter(Boolean)`). Other opts: `label` (display), `phase` (progress group), and independent `provider`/`model` LLM target overrides (either may be provided alone). Anything else (`effort`/`isolation`/`agentType`) is rejected loudly. - `pipeline(items, ...stages): Promise` — run each item through the stages independently with NO barrier between stages (prefer this for multi-stage work). Each stage receives `(prev, item, index)`. An ordinary stage throw drops that ITEM to `null` and skips its remaining stages. - `parallel(thunks): Promise` — run zero-argument functions concurrently and await ALL of them (a barrier; use only when a stage genuinely needs every prior result together). A throwing thunk resolves to `null`. - `phase(title)` — start a progress phase; `log(message)` — narrate progress; `args` — the tool call's `args` input, verbatim. Misused hooks (bad arguments, unknown options, unsupported schemas, tripped caps) throw errors that ALWAYS kill the script — they never dissolve into a per-item `null`. Constraints: concurrency and total-agent caps apply; no filesystem, network, timers, or Node.js APIs are provided — the agents do the work, the script only coordinates them. The run executes in the foreground: this call returns when the whole script finishes. */\n workflow: {\n /** The plain-JS workflow script body (top-level await allowed; NO `export const meta` statement; end with `return `). */\n script: string;\n /** The workflow identity block (plain JSON — never code). */\n meta: {\n /** Short kebab-case workflow name. */\n name: string;\n /** One-line description of what the workflow does. */\n description: string;\n /** Optional guidance on when this workflow applies. */\n whenToUse?: string;\n /** Optional phase declarations matched by phase() calls. */\n phases?: ({\n /** The phase title phase() calls match by exact string. */\n title: string;\n /** Optional one-line description of the phase. */\n detail?: string;\n /** Optional provider override this phase is expected to use. */\n provider?: string;\n /** Optional model override this phase is expected to use. */\n model?: string;\n } & Record)[];\n } & Record;\n /** Optional JSON input exposed to the script as the `args` global (wrap a bare list as a field, e.g. {\"files\": [...]}). */\n args?: Record;\n } & Record;\n /** Create or fully replace a UTF-8 text file. */\n write: {\n /** Path to write, resolved by the filesystem backend. */\n file_path: string;\n /** Full UTF-8 text content to write. */\n content: string;\n } & Record;\n}\n\ninterface ToolOutputMap {\n bash: {\n kind: \"background\";\n taskId: string;\n } | {\n kind: \"foreground\";\n exitCode: number | null;\n signal: string | null;\n timedOut: boolean;\n aborted: boolean;\n timeoutMs: number;\n stdout: {\n text: string;\n truncated: boolean;\n spillPath?: string;\n };\n stderr: {\n text: string;\n truncated: boolean;\n spillPath?: string;\n };\n sandbox?: {\n mode: string;\n denied: boolean;\n enforcement?: string;\n runnerFailed?: boolean;\n };\n };\n cordis_inspect: string;\n cordis_mount: {\n id: string;\n pluginName: string;\n state: \"pending\" | \"loading\" | \"active\" | \"failed\" | \"disposed\" | \"unloading\";\n provides: string[];\n waitingFor: string[];\n };\n cordis_unmount: {\n id: string;\n pluginName: string;\n };\n edit: {\n path: string;\n before: string;\n after: string;\n };\n ralph: {\n runId: string;\n agentsStarted: number;\n result: JsonValue;\n };\n read: {\n path: string;\n offset: number;\n lines: {\n number: number;\n text: string;\n }[];\n totalLines: number;\n };\n skill: {\n name: string;\n provider: string;\n resourceBase?: {\n kind: \"directory\";\n path: string;\n } | {\n kind: \"url\";\n url: string;\n } | {\n kind: \"opaque\";\n description: string;\n };\n content: string;\n };\n subagent: {\n kind: \"background\";\n taskId: string;\n } | {\n kind: \"foreground\";\n runId: string;\n output: JsonValue[];\n };\n subagent_fork: {\n kind: \"background\";\n taskId: string;\n } | {\n kind: \"foreground\";\n runId: string;\n output: JsonValue[];\n };\n task_kill: {\n outcome: \"cancellation-requested\" | \"already-finished\";\n task: {\n id: string;\n kind: string;\n label: string;\n status: \"running\" | \"stopping\" | \"completed\" | \"killed\" | \"failed\";\n detail?: string;\n startedAt: number;\n finishedAt?: number;\n };\n };\n task_list: ({\n id: string;\n kind: string;\n label: string;\n status: \"running\" | \"stopping\" | \"completed\" | \"killed\" | \"failed\";\n detail?: string;\n startedAt: number;\n finishedAt?: number;\n })[];\n task_output: {\n text: string;\n task: {\n id: string;\n kind: string;\n label: string;\n status: \"running\" | \"stopping\" | \"completed\" | \"killed\" | \"failed\";\n detail?: string;\n startedAt: number;\n finishedAt?: number;\n };\n };\n todo_write: {\n todos: ({\n content: string;\n status: \"pending\" | \"in_progress\" | \"completed\";\n })[];\n counts: {\n pending: number;\n inProgress: number;\n completed: number;\n };\n };\n workflow: {\n runId: string;\n agentsStarted: number;\n result: JsonValue;\n };\n write: {\n path: string;\n operation: \"create\" | \"update\";\n before: string | null;\n after: string;\n };\n}\n\ntype ToolName = keyof ToolOutputMap\n\ndeclare class ToolCallError extends Error {\n readonly name: \"ToolCallError\";\n readonly toolName: ToolName;\n}\n\ndeclare const tools: {\n [K in ToolName]: (args: ToolArgsMap[K]) => Promise;\n}\n```","tools":[{"name":"bash","description":"Execute a bash command (`bash -c`) and return its stdout/stderr. Each call runs in a fresh shell: no state (cwd, variables, functions) persists between calls — pass `workdir` instead of using `cd`. Non-zero exits are reported as `[exit code: N]`. Current harness environment facts are exposed through managed `$DSH_*` variables; inspect them when needed. Commands may run under a file sandbox; a blocked file operation is reported as `[sandbox: file access denied under mode]` — a policy denial, not a bug in the command; do not retry another way. Long output is truncated to its tail; the full output is saved to a file whose path is reported when available. Set `run_in_background: true` for long-running commands: the call returns a task id immediately; read its output with `task_output` and stop it with `task_kill`.","parameters":{"type":"object","properties":{"command":{"type":"string","description":"The bash command to execute."},"description":{"type":"string","description":"Clear, concise description of what this command does in active voice, 5-10 words (shown in the UI). Examples: \"ls\" → \"List files in current directory\"; \"git status\" → \"Show working tree status\"; \"npm install\" → \"Install package dependencies\"."},"timeoutMs":{"type":"number","description":"Timeout in milliseconds. The executor applies its configured default and cap, and kills the command on expiry."},"workdir":{"type":"string","description":"Working directory for this command. Defaults to the session workspace; a relative path is resolved against it."},"run_in_background":{"type":"boolean","description":"Run in the background and return a task id immediately (collect with task_output, stop with task_kill). No timeout applies."}},"required":["command","description"]}},{"name":"cordis_inspect","description":"Inspect the live Cordis runtime in the current DSH process. Read-only. Sections: `services` (every provided ctx service and the plugin fiber that owns it), `plugins` (all live plugin fibers with their lifecycle states), `tools` (the model-facing tools currently registered, i.e. what you can call), `temporary` (only temporary Plugins created by cordis_mount: id, name, state, provided services, awaited services, and lifetime), `api` (method signatures AND argument/return type shapes for every LIVE service — read this before writing plugin code that calls a service), `events` (every harness event with its dispatch mode and exact signature — pick listener targets here). Temporary Plugins exist only in memory, remain active across later turns, and disappear after cordis_unmount, toolset unload, or DSH restart; they are not restored automatically. The `temporary` section is a subset of `plugins`. Omit `what` to get all six sections. With `what:\"api\"` or `what:\"events\"`, pass an exact `name` to narrow to one service/event and include its original source JSDoc.","parameters":{"type":"object","properties":{"what":{"type":"string","description":"Limit the report to one section. Omit for all sections.","enum":["services","plugins","tools","temporary","api","events"]},"name":{"type":"string","description":"Exact service key or event name whose original JSDoc to include; valid only with what:\"api\" or what:\"events\"."}}}},{"name":"cordis_mount","description":"Mount a temporary Cordis Plugin in the current DSH process. This creates an in-memory runtime Plugin, not an installed or configured Plugin. It remains active across later turns until cordis_unmount, toolset unload, or DSH restart. It does not create files, install a package, change cordis.yml or personal/project config, survive restart, or automatically become permanent. To keep it, ask the Agent to implement a normal local, project, or repository Plugin through the regular development workflow. It may affect other sessions in the same process; the sandbox is not a security boundary, and injected services reach the real runtime. `code` runs now as the body of an async JavaScript function in an isolated sandbox and MUST `return` a plugin. Two forms: FUNCTION form `return (ctx) => { … }` — declares no inject, so it can register tools, listen to events, and provide services, but reaching ANY service (e.g. ctx.bash) throws; use it only when you need no services. OBJECT form `return { name?, inject: ['bash', 'llm', …], apply(ctx) { … } }` — declares dependencies, and cordis activates the plugin only after the services exist; PREFER this form. You may reach ONLY the services you list in inject: an undeclared service throws even if it exists, because an undeclared dependency would not be cleaned up if its provider is unmounted. BEFORE calling a service from your code, read cordis_inspect what:\"api\" — it lists method signatures AND the type shapes of their arguments/returns (do not guess a field's type; e.g. a bash run's stdout is an object, not a string). Inside `apply`, use the standard cordis API: `ctx.on(event, listener)` to observe events (see cordis_inspect what:\"events\"), or call `harness.registerTool(ctx, harness.defineTool({ name, description, parameters: { text: { type: 'string', required: true } }, output: { schema: { type: 'string' }, render(_args, value) { return [{ type: 'text', text: value }] } }, async execute(args) { return args.text } }))` to give yourself a new tool — it becomes callable on your NEXT step. Tool parameters: each key IS a property — { type: 'string'|'number'|'integer'|'boolean'|'null'|'object'|'array'|'json', required?: true, description?, enum?, const?, items?, properties? }; every direct DSL object declares additionalProperties: true|false, and oneOf: [schema, schema, ...] replaces type for an exact-one union. A raw JSON-Schema { type: 'object', properties, required?: […] } wrapper is also accepted with open-by-default objects. A tool's `execute` MUST return the lossless JSON value declared by `output.schema`; `output.render(args, value)` separately returns Native/model content blocks. Temporary Plugins can COMPOSE: one Plugin may `ctx.provide('name', value)` a service and another may declare `inject: ['name']` to consume it — the consumer stays pending until the provider exists and returns to pending when the provider is unmounted. Everything registered inside `apply` is cleaned up automatically by cordis_unmount. Sandbox globals: `console` (tagged `[cordis:]`, writes through to the harness terminal), `harness.defineTool`, `harness.registerTool`, `btoa`, `atob`, `TextEncoder`, `TextDecoder`. Node APIs are DISABLED — do filesystem/network/timer work through the cordis services, never Node built-ins: `require`, `setTimeout`/`setInterval`, and `fetch` throw redirect errors; `process` and `Buffer` are undefined. Instead use inject: ['fs'] + ctx.fs for files, inject: ['web'] + ctx.web for HTTP, inject: ['bash'] + ctx.bash for processes, and inject: ['timer'] + ctx.setTimeout/ctx.setInterval for timing (fiber effects, auto-cleaned when unmounted) — cordis_inspect what:\"api\" shows what THIS runtime provides. Write PLAIN JavaScript, not TypeScript (no `as`, no type annotations). Cautions: (1) waterfall events (e.g. tools/pre-execute) hand the listener a trailing `next` callback which MUST be called — returning without `next()` VETOES the call; prefer plain notification events unless you intend to intercept. (2) Never await something that only resolves after the current turn (your code runs INSIDE a tool call of that turn — it would deadlock). (3) Your `ctx` is a restricted façade: you can register tools, observe events, provide/consume services, and use timers, but framework internals (ctx.root, ctx.fiber, ctx.extend, ctx.plugin, …) are withheld. It is not a security boundary though — the services you inject (e.g. ctx.bash) reach the real runtime.","parameters":{"type":"object","properties":{"code":{"type":"string","description":"JavaScript body returning a temporary Plugin; evaluated now and saved nowhere."}},"required":["code"]}},{"name":"cordis_unmount","description":"Unmount a current-process temporary Plugin created by cordis_mount. Waits for its tools, listeners, services, timers, and other owned effects to clean up completely. Only dyn-N temporary ids are accepted; this cannot remove Loader, configured, or installed Plugins.","parameters":{"type":"object","properties":{"id":{"type":"string","description":"The temporary Plugin id returned by cordis_mount (for example \"dyn-1\"); valid only in this process and invalid after unmount or restart."}},"required":["id"]}},{"name":"edit","description":"Edit an existing UTF-8 text file by replacing literal text.","parameters":{"type":"object","properties":{"file_path":{"type":"string","description":"Path to edit, resolved by the filesystem backend."},"old_string":{"type":"string","description":"Literal text to replace. Must match exactly."},"new_string":{"type":"string","description":"Literal replacement text. Use an empty string to delete the match."},"replace_all":{"type":"boolean","description":"Replace all matches. Defaults to false; when false, old_string must appear exactly once."}},"required":["file_path","old_string","new_string"]}},{"name":"ralph","description":"Run a foreground fresh-agent Ralph loop toward one immutable objective. Use only when the direct human explicitly asks for Ralph or fresh-agent iteration. Each round opens a new child with no parent conversation or prior child session; the shared workspace is long-term memory, and only a bounded structured report crosses rounds. The call returns when a worker reports completion or a concrete blocker, or at the round limit. Ordinary long-running same-session work belongs to goal tools.","parameters":{"type":"object","properties":{"objective":{"type":"string","description":"The immutable completion objective for every fresh Ralph round."},"maxRounds":{"type":"number","description":"Optional positive safe-integer round cap, bounded by the deployment ceiling."}},"required":["objective"]}},{"name":"read","description":"Read a UTF-8 text file and return line-numbered content.","parameters":{"type":"object","properties":{"file_path":{"type":"string","description":"Path to read, resolved by the filesystem backend."},"offset":{"type":"number","description":"1-based first line to return. Defaults to 1."},"limit":{"type":"number","description":"Maximum number of lines to return. Defaults to 2000."}},"required":["file_path"]}},{"name":"run_code","description":"Execute a TypeScript program against the available tools. Write the BODY of an async function (erasable syntax only; top-level `await` and `return` work) and call tools as `await tools.name(args)` per the declarations in the system prompt. Only what you print or return comes back — curate it.","parameters":{"type":"object","properties":{"code":{"type":"string","description":"The program: the body of an async TypeScript function."},"description":{"type":"string","description":"Clear, concise description of what this program does in active voice, 5-10 words (shown in the UI). Examples: \"Count TODO markers across packages\"; \"Read failing test and its fixture\"; \"Rename config key in every cordis.yml\"."}},"required":["code","description"]}},{"name":"skill","description":"Load the full instructions for an available skill. Call this with the exact skill name from the session skill catalog before acting on a task that names or clearly matches that skill.","parameters":{"type":"object","properties":{"name":{"type":"string","description":"The exact skill name from the available skills list."}},"required":["name"]}},{"name":"subagent","description":"Delegate a self-contained task to a subagent (a separate agent that works in its own context) and return its final result. Use this to offload focused, independent work — research, a scoped implementation, an analysis — so it does not consume this conversation's context. The subagent runs to completion and you receive only its final answer, not its intermediate steps. Give it a complete, standalone prompt: it does not see this conversation. Set `run_in_background: true` to return a task id; collect with `task_output` and stop with `task_kill`.","parameters":{"type":"object","properties":{"description":{"type":"string","description":"A short (3-5 word) description of the delegated task, for display."},"prompt":{"type":"string","description":"The complete, self-contained task for the subagent. It does not share this conversation's context, so include everything it needs."},"run_in_background":{"type":"boolean","description":"Run as a background task and return its id; collect with task_output or stop with task_kill."}},"required":["description","prompt"]}},{"name":"subagent_fork","description":"Delegate a task to a subagent that inherits this conversation: a child agent seeded with all completed turns so far (it does not see the current in-flight turn), returning only its final result. Use this when the subtask builds on this conversation's context — a follow-up analysis, a review, a continuation — without consuming this conversation's context for the work itself. You receive only its final answer, not its intermediate steps. Set `run_in_background: true` to return a task id; collect with `task_output` and stop with `task_kill`.","parameters":{"type":"object","properties":{"description":{"type":"string","description":"A short (3-5 word) description of the delegated task, for display."},"prompt":{"type":"string","description":"The task for the subagent. It already sees this conversation's completed turns, so build on them freely and state only what is new."},"run_in_background":{"type":"boolean","description":"Run as a background task and return its id; collect with task_output or stop with task_kill."}},"required":["description","prompt"]}},{"name":"task_kill","description":"Request cancellation of a running background task by task id. Returns immediately; the task settles as killed once its work actually stops.","parameters":{"type":"object","properties":{"task_id":{"type":"string","description":"Task id returned by the tool that started the background work."},"reason":{"type":"string","description":"Optional short reason, recorded in the log and forwarded to the task."}},"required":["task_id"]}},{"name":"task_list","description":"List your background tasks (running and finished) with their ids, kinds, and statuses.","parameters":{"type":"object","properties":{}}},{"name":"task_output","description":"Read a background task. Stream tasks return only output since the previous read; final-output tasks return their result after settlement. Every response ends with `[status: ...]`. Reads are non-blocking unless `wait: true`, which waits up to the configured cap.","parameters":{"type":"object","properties":{"task_id":{"type":"string","description":"Task id returned by the tool that started the background work."},"wait":{"type":"boolean","description":"Block until the task reaches a terminal status or the timeout expires. A timed-out wait returns [status: running] and leaves the task alive."},"timeout_ms":{"type":"number","description":"Max wait in milliseconds (only meaningful with wait: true). Defaults to the configured wait timeout; capped by the configured maximum."}},"required":["task_id"]}},{"name":"todo_write","description":"Record and update a structured task list for the current work. Send the ENTIRE list every call — it REPLACES the previous list (there are no partial updates, no per-item edits). Use it to plan multi-step work and show progress: add one todo per concrete step before you start. Keep AT MOST ONE todo `in_progress` at a time; while work remains, exactly one active task should be `in_progress`. Mark a todo `completed` the moment it is done (do not batch completions), and allow no `in_progress` item only once all work is complete. Skip the list for trivial single-step tasks. Statuses: `pending` (not started), `in_progress` (being worked on now), `completed` (finished).","parameters":{"type":"object","properties":{"todos":{"type":"array","description":"The COMPLETE task list, replacing any previous list.","items":{"type":"object","additionalProperties":false,"properties":{"content":{"type":"string","description":"What the task is — a short imperative line."},"status":{"type":"string","description":"pending (not started) | in_progress (now) | completed (done).","enum":["pending","in_progress","completed"]}},"required":["content","status"]}}},"required":["todos"]}},{"name":"workflow","description":"Run a JavaScript workflow script that orchestrates subagents at scale. Use this for work that fans out across many independent pieces — an audit over many files, a migration, multi-angle research, adversarial verification of findings — where you write the orchestration as a script instead of delegating turn by turn.\n\nThe workflow's identity rides the `meta` parameter as JSON: required `name` (short kebab-case) and `description` strings, optional `whenToUse` string and `phases` array (`{title, detail?, provider?, model?}`). The `script` parameter is the plain JavaScript body ONLY (NOT TypeScript, and NO `export const meta` statement — meta is a parameter, not code), running with top-level await; end with `return ` — the value must be JSON-serializable and is this tool's result.\n\nScript-body hooks:\n- `agent(prompt, opts?): Promise` — run one subagent to completion. Without `opts.schema` it resolves to the child's final text; with `opts.schema` (an object-rooted JSON Schema using ONLY type/properties/required/additionalProperties/items/enum/const/oneOf — no pattern/format/numeric bounds) it resolves to the validated object. Resolves `null` when the child fails (filter with `.filter(Boolean)`). Other opts: `label` (display), `phase` (progress group), and independent `provider`/`model` LLM target overrides (either may be provided alone). Anything else (`effort`/`isolation`/`agentType`) is rejected loudly.\n- `pipeline(items, ...stages): Promise` — run each item through the stages independently with NO barrier between stages (prefer this for multi-stage work). Each stage receives `(prev, item, index)`. An ordinary stage throw drops that ITEM to `null` and skips its remaining stages.\n- `parallel(thunks): Promise` — run zero-argument functions concurrently and await ALL of them (a barrier; use only when a stage genuinely needs every prior result together). A throwing thunk resolves to `null`.\n- `phase(title)` — start a progress phase; `log(message)` — narrate progress; `args` — the tool call's `args` input, verbatim.\n\nMisused hooks (bad arguments, unknown options, unsupported schemas, tripped caps) throw errors that ALWAYS kill the script — they never dissolve into a per-item `null`.\n\nConstraints: concurrency and total-agent caps apply; no filesystem, network, timers, or Node.js APIs are provided — the agents do the work, the script only coordinates them. The run executes in the foreground: this call returns when the whole script finishes.","parameters":{"type":"object","properties":{"script":{"type":"string","description":"The plain-JS workflow script body (top-level await allowed; NO `export const meta` statement; end with `return `)."},"meta":{"type":"object","description":"The workflow identity block (plain JSON — never code).","additionalProperties":true,"properties":{"name":{"type":"string","description":"Short kebab-case workflow name."},"description":{"type":"string","description":"One-line description of what the workflow does."},"whenToUse":{"type":"string","description":"Optional guidance on when this workflow applies."},"phases":{"type":"array","description":"Optional phase declarations matched by phase() calls.","items":{"type":"object","additionalProperties":true,"properties":{"title":{"type":"string","description":"The phase title phase() calls match by exact string."},"detail":{"type":"string","description":"Optional one-line description of the phase."},"provider":{"type":"string","description":"Optional provider override this phase is expected to use."},"model":{"type":"string","description":"Optional model override this phase is expected to use."}},"required":["title"]}}},"required":["name","description"]},"args":{"type":"object","description":"Optional JSON input exposed to the script as the `args` global (wrap a bare list as a field, e.g. {\"files\": [...]}).","additionalProperties":true}},"required":["script","meta"]}},{"name":"write","description":"Create or fully replace a UTF-8 text file.","parameters":{"type":"object","properties":{"file_path":{"type":"string","description":"Path to write, resolved by the filesystem backend."},"content":{"type":"string","description":"Full UTF-8 text content to write."}},"required":["file_path","content"]}}]},"reason":"initial"}} {"type":"assistant/chunk","seq":5,"time":1783950000005,"data":{"turn":1,"step":1,"chunk":{"type":"block-start","index":0,"blockType":"tool-call"}}} {"type":"assistant/chunk","seq":6,"time":1783950000006,"data":{"turn":1,"step":1,"chunk":{"type":"tool-call-delta","index":0,"id":"advanced-mount","name":"cordis_mount","argumentsDelta":"{\"code\":\"return { name: 'snapshot-marker', apply() {} }\"}"}}} {"type":"assistant/chunk","seq":7,"time":1783950000007,"data":{"turn":1,"step":1,"chunk":{"type":"block-end","index":0,"block":{"type":"tool-call","id":"advanced-mount","name":"cordis_mount","arguments":"{\"code\":\"return { name: 'snapshot-marker', apply() {} }\"}"}}}} diff --git a/examples/headless-agent/tests/snapshots/ralph-loop/session.1.jsonl b/examples/headless-agent/tests/snapshots/ralph-loop/session.1.jsonl index ca36330c36..d684022863 100644 --- a/examples/headless-agent/tests/snapshots/ralph-loop/session.1.jsonl +++ b/examples/headless-agent/tests/snapshots/ralph-loop/session.1.jsonl @@ -1,4 +1,4 @@ -{"type":"session","version":0,"id":"42222222-2222-4222-8222-222222222222","createdAt":1783951001000,"cwd":"/tmp/ralph-headless","parentSession":"41111111-1111-4111-8111-111111111111"} +{"type":"session","version":0,"id":"42222222-2222-4222-8222-222222222222","createdAt":1783951001000,"cwd":"{{cwd}}","parentSession":"41111111-1111-4111-8111-111111111111"} {"type":"assistant/chunk","seq":0,"time":1783951001001,"data":{"turn":1,"step":1,"chunk":{"type":"block-start","index":0,"blockType":"tool-call"}}} {"type":"assistant/chunk","seq":1,"time":1783951001002,"data":{"turn":1,"step":1,"chunk":{"type":"tool-call-delta","index":0,"id":"round-one-report","name":"structured_output","argumentsDelta":"{\"status\":\"continue\",\"summary\":\"ROUND_ONE_HANDOFF\",\"evidence\":[\"Round one inspected the workspace.\"],\"nextSteps\":[\"Finish the snapshot objective.\"],\"blocker\":\"\"}"}}} {"type":"assistant/chunk","seq":2,"time":1783951001003,"data":{"turn":1,"step":1,"chunk":{"type":"block-end","index":0,"block":{"type":"tool-call","id":"round-one-report","name":"structured_output","arguments":"{\"status\":\"continue\",\"summary\":\"ROUND_ONE_HANDOFF\",\"evidence\":[\"Round one inspected the workspace.\"],\"nextSteps\":[\"Finish the snapshot objective.\"],\"blocker\":\"\"}"}}}} diff --git a/examples/headless-agent/tests/snapshots/ralph-loop/session.2.jsonl b/examples/headless-agent/tests/snapshots/ralph-loop/session.2.jsonl index c722158098..15f54b231b 100644 --- a/examples/headless-agent/tests/snapshots/ralph-loop/session.2.jsonl +++ b/examples/headless-agent/tests/snapshots/ralph-loop/session.2.jsonl @@ -1,4 +1,4 @@ -{"type":"session","version":0,"id":"43333333-3333-4333-8333-333333333333","createdAt":1783951002000,"cwd":"/tmp/ralph-headless","parentSession":"41111111-1111-4111-8111-111111111111"} +{"type":"session","version":0,"id":"43333333-3333-4333-8333-333333333333","createdAt":1783951002000,"cwd":"{{cwd}}","parentSession":"41111111-1111-4111-8111-111111111111"} {"type":"assistant/chunk","seq":0,"time":1783951002001,"data":{"turn":1,"step":1,"chunk":{"type":"block-start","index":0,"blockType":"tool-call"}}} {"type":"assistant/chunk","seq":1,"time":1783951002002,"data":{"turn":1,"step":1,"chunk":{"type":"tool-call-delta","index":0,"id":"round-two-report","name":"structured_output","argumentsDelta":"{\"status\":\"complete\",\"summary\":\"The Ralph snapshot objective is complete.\",\"evidence\":[\"Two fresh rounds completed through the shipped app.\"],\"nextSteps\":[],\"blocker\":\"\"}"}}} {"type":"assistant/chunk","seq":2,"time":1783951002003,"data":{"turn":1,"step":1,"chunk":{"type":"block-end","index":0,"block":{"type":"tool-call","id":"round-two-report","name":"structured_output","arguments":"{\"status\":\"complete\",\"summary\":\"The Ralph snapshot objective is complete.\",\"evidence\":[\"Two fresh rounds completed through the shipped app.\"],\"nextSteps\":[],\"blocker\":\"\"}"}}}} diff --git a/examples/headless-agent/tests/snapshots/ralph-loop/session.jsonl b/examples/headless-agent/tests/snapshots/ralph-loop/session.jsonl index c452fb5fa8..4b6fe2dcb5 100644 --- a/examples/headless-agent/tests/snapshots/ralph-loop/session.jsonl +++ b/examples/headless-agent/tests/snapshots/ralph-loop/session.jsonl @@ -1 +1 @@ -{"type":"session","version":0,"id":"41111111-1111-4111-8111-111111111111","createdAt":1783951000000,"cwd":"/tmp/ralph-headless"} +{"type":"session","version":0,"id":"41111111-1111-4111-8111-111111111111","createdAt":1783951000000,"cwd":"{{cwd}}"} diff --git a/examples/headless-agent/tests/subagent-inheritance-snapshots/parent-override/child.expected.jsonl b/examples/headless-agent/tests/subagent-inheritance-snapshots/parent-override/child.expected.jsonl new file mode 100644 index 0000000000..59a93af0f6 --- /dev/null +++ b/examples/headless-agent/tests/subagent-inheritance-snapshots/parent-override/child.expected.jsonl @@ -0,0 +1,25 @@ +{"type":"session","version":0,"id":"{{sessionId}}","createdAt":0,"cwd":"{{cwd}}","parentSession":"{{sessionId}}","delegationDepth":1} +{"type":"sandbox/mode","seq":0,"time":0,"data":{"mode":"read-only","source":"delegation"}} +{"type":"turn/start","seq":1,"time":0,"data":{"turn":1,"trigger":{"kind":"message","source":{"kind":"user"}}}} +{"type":"user/message","seq":2,"time":0,"data":{"content":[{"type":"text","text":"Use the write tool exactly once with file_path set to exactly the relative path inherited.txt and content escaped. If the write is denied, reply with the single word CHILD_DENIED and the denial marker line; do not retry and do not request escalation. If it succeeds, reply CHILD_WROTE."}],"source":{"kind":"user"},"role":"user","id":"{{sessionId}}"},"surfaceOp":"append"} +{"type":"session/title","seq":3,"time":0,"data":{"title":"Use the write tool exactly","messageSeqs":[2],"source":{"kind":"fallback"}}} +{"type":"step/start","seq":4,"time":0,"data":{"turn":1,"step":1}} +{"type":"request/header","seq":5,"time":0,"data":{"header":{"config":{"provider":"deepseek","model":"deepseek-v4-flash"},"system":"{{system}}","tools":"{{tools}}"},"reason":"initial"}} +{"type":"assistant/chunk","seq":6,"time":0,"data":{"turn":1,"step":1,"chunk":{"type":"block-start","index":0,"blockType":"tool-call"}}} +{"type":"assistant/chunk","seq":7,"time":0,"data":{"turn":1,"step":1,"chunk":{"type":"tool-call-delta","index":0,"id":"child-write","name":"write","argumentsDelta":"{\"file_path\": \"inherited.txt\", \"content\": \"escaped\"}"}}} +{"type":"assistant/chunk","seq":8,"time":0,"data":{"turn":1,"step":1,"chunk":{"type":"block-end","index":0,"block":{"type":"tool-call","id":"child-write","name":"write","arguments":"{\"file_path\": \"inherited.txt\", \"content\": \"escaped\"}"}}}} +{"type":"assistant/chunk","seq":9,"time":0,"data":{"turn":1,"step":1,"chunk":{"type":"usage","usage":{"inputTokens":10,"outputTokens":5}}}} +{"type":"assistant/chunk","seq":10,"time":0,"data":{"turn":1,"step":1,"chunk":{"type":"finish","reason":{"kind":"tool-calls"}}}} +{"type":"assistant/message","seq":11,"time":0,"data":{"turn":1,"step":1,"message":{"role":"assistant","content":[{"type":"tool-call","id":"child-write","name":"write","arguments":"{\"file_path\": \"inherited.txt\", \"content\": \"escaped\"}"}],"source":{"kind":"model","provider":"deepseek","model":"deepseek-v4-flash"},"id":"{{sessionId}}"},"usage":{"inputTokens":10,"outputTokens":5}},"sourceEventSeqs":[6,7,8,9,10],"surfaceOp":"append"} +{"type":"tool/call","seq":12,"time":0,"data":{"turn":1,"step":1,"callId":"child-write","name":"write","arguments":"{\"file_path\": \"inherited.txt\", \"content\": \"escaped\"}"}} +{"type":"tool/result","seq":13,"time":0,"data":{"turn":1,"step":1,"message":{"source":{"kind":"tool","callId":"child-write"},"content":[{"type":"tool-result","toolCallId":"child-write","content":[{"type":"text","text":"Error: [sandbox: file access denied under read-only mode]\n[sandbox: escalation available — retry this exact operation once with sandbox_permissions (the narrowest wider mode that suffices) + justification; the approval prompt asks the user]"}],"isError":true}],"role":"user","id":"{{sessionId}}"},"error":{"name":"FsError","code":"FS_SANDBOX_DENIED"}},"sourceEventSeqs":[12],"surfaceOp":"append"} +{"type":"step/end","seq":14,"time":0,"data":{"turn":1,"step":1}} +{"type":"step/start","seq":15,"time":0,"data":{"turn":1,"step":2}} +{"type":"assistant/chunk","seq":16,"time":0,"data":{"turn":1,"step":2,"chunk":{"type":"block-start","index":0,"blockType":"text"}}} +{"type":"assistant/chunk","seq":17,"time":0,"data":{"turn":1,"step":2,"chunk":{"type":"text-delta","index":0,"text":"CHILD_DENIED [sandbox: file access denied under read-only mode]"}}} +{"type":"assistant/chunk","seq":18,"time":0,"data":{"turn":1,"step":2,"chunk":{"type":"block-end","index":0,"block":{"type":"text","text":"CHILD_DENIED [sandbox: file access denied under read-only mode]"}}}} +{"type":"assistant/chunk","seq":19,"time":0,"data":{"turn":1,"step":2,"chunk":{"type":"usage","usage":{"inputTokens":10,"outputTokens":5}}}} +{"type":"assistant/chunk","seq":20,"time":0,"data":{"turn":1,"step":2,"chunk":{"type":"finish","reason":{"kind":"stop"}}}} +{"type":"assistant/message","seq":21,"time":0,"data":{"turn":1,"step":2,"message":{"role":"assistant","content":[{"type":"text","text":"CHILD_DENIED [sandbox: file access denied under read-only mode]"}],"source":{"kind":"model","provider":"deepseek","model":"deepseek-v4-flash"},"id":"{{sessionId}}"},"usage":{"inputTokens":10,"outputTokens":5}},"sourceEventSeqs":[16,17,18,19,20],"surfaceOp":"append"} +{"type":"step/end","seq":22,"time":0,"data":{"turn":1,"step":2}} +{"type":"turn/end","seq":23,"time":0,"data":{"turn":1,"reason":{"kind":"completed"}}} diff --git a/examples/headless-agent/tests/subagent-inheritance-snapshots/parent-override/child.replay.jsonl b/examples/headless-agent/tests/subagent-inheritance-snapshots/parent-override/child.replay.jsonl new file mode 100644 index 0000000000..99947a3797 --- /dev/null +++ b/examples/headless-agent/tests/subagent-inheritance-snapshots/parent-override/child.replay.jsonl @@ -0,0 +1,18 @@ +{"type": "session", "version": 0, "id": "subagent-inheritance-child", "createdAt": 2, "delegationDepth": 1} +{"type":"turn/start","seq":0,"time":1,"data":{"turn":1,"trigger":{"kind":"message","source":{"kind":"user"}}}} +{"type":"user/message","seq":1,"time":2,"data":{"content":[{"type":"text","text":"delegated task"}],"source":{"kind":"user"}},"surfaceOp":"append"} +{"type":"step/start","seq":2,"time":3,"data":{"turn":1,"step":1}} +{"type":"assistant/chunk","seq":3,"time":4,"data":{"turn":1,"step":1,"chunk":{"type":"block-start","index":0,"blockType":"tool-call"}}} +{"type":"assistant/chunk","seq":4,"time":5,"data":{"turn":1,"step":1,"chunk":{"type":"tool-call-delta","index":0,"id":"child-write","name":"write","argumentsDelta":"{\"file_path\": \"inherited.txt\", \"content\": \"escaped\"}"}}} +{"type":"assistant/chunk","seq":5,"time":6,"data":{"turn":1,"step":1,"chunk":{"type":"block-end","index":0,"block":{"type":"tool-call","id":"child-write","name":"write","arguments":"{\"file_path\": \"inherited.txt\", \"content\": \"escaped\"}"}}}} +{"type":"assistant/chunk","seq":6,"time":7,"data":{"turn":1,"step":1,"chunk":{"type":"usage","usage":{"inputTokens":10,"outputTokens":5}}}} +{"type":"assistant/chunk","seq":7,"time":8,"data":{"turn":1,"step":1,"chunk":{"type":"finish","reason":{"kind":"tool-calls"}}}} +{"type":"step/end","seq":8,"time":9,"data":{"turn":1,"step":1}} +{"type":"step/start","seq":9,"time":10,"data":{"turn":1,"step":2}} +{"type":"assistant/chunk","seq":10,"time":11,"data":{"turn":1,"step":2,"chunk":{"type":"block-start","index":0,"blockType":"text"}}} +{"type":"assistant/chunk","seq":11,"time":12,"data":{"turn":1,"step":2,"chunk":{"type":"text-delta","index":0,"text":"CHILD_DENIED [sandbox: file access denied under read-only mode]"}}} +{"type":"assistant/chunk","seq":12,"time":13,"data":{"turn":1,"step":2,"chunk":{"type":"block-end","index":0,"block":{"type":"text","text":"CHILD_DENIED [sandbox: file access denied under read-only mode]"}}}} +{"type":"assistant/chunk","seq":13,"time":14,"data":{"turn":1,"step":2,"chunk":{"type":"usage","usage":{"inputTokens":10,"outputTokens":5}}}} +{"type":"assistant/chunk","seq":14,"time":15,"data":{"turn":1,"step":2,"chunk":{"type":"finish","reason":{"kind":"stop"}}}} +{"type":"step/end","seq":15,"time":16,"data":{"turn":1,"step":2}} +{"type":"turn/end","seq":16,"time":17,"data":{"turn":1,"reason":{"kind":"completed"}}} diff --git a/examples/headless-agent/tests/subagent-inheritance-snapshots/parent-override/parent.expected.jsonl b/examples/headless-agent/tests/subagent-inheritance-snapshots/parent-override/parent.expected.jsonl new file mode 100644 index 0000000000..d5bfb405de --- /dev/null +++ b/examples/headless-agent/tests/subagent-inheritance-snapshots/parent-override/parent.expected.jsonl @@ -0,0 +1,28 @@ +{"type":"session","version":0,"id":"{{sessionId}}","createdAt":0,"cwd":"{{cwd}}","delegationDepth":0} +{"type":"turn/start","seq":0,"time":0,"data":{"turn":1,"trigger":{"kind":"message","source":{"kind":"user"}}}} +{"type":"user/message","seq":1,"time":0,"data":{"content":[{"type":"text","text":"Tighten this session to read-only."}],"source":{"kind":"user"},"role":"user","id":"{{sessionId}}"},"surfaceOp":"append"} +{"type":"sandbox/mode","seq":2,"time":0,"data":{"mode":"read-only"}} +{"type":"turn/end","seq":3,"time":0,"data":{"turn":1,"reason":{"kind":"completed"}}} +{"type":"turn/start","seq":4,"time":0,"data":{"turn":2,"trigger":{"kind":"message","source":{"kind":"user"}}}} +{"type":"user/message","seq":5,"time":0,"data":{"content":[{"type":"text","text":"Delegate the write probe to a subagent."}],"source":{"kind":"user"},"role":"user","id":"{{sessionId}}"},"surfaceOp":"append"} +{"type":"session/title","seq":6,"time":0,"data":{"title":"Tighten this session to read-only.","messageSeqs":[1],"source":{"kind":"fallback"}}} +{"type":"step/start","seq":7,"time":0,"data":{"turn":2,"step":1}} +{"type":"request/header","seq":8,"time":0,"data":{"header":{"config":{"provider":"deepseek","model":"deepseek-v4-flash"},"system":"{{system}}","tools":"{{tools}}"},"reason":"initial"}} +{"type":"assistant/chunk","seq":9,"time":0,"data":{"turn":2,"step":1,"chunk":{"type":"block-start","index":0,"blockType":"tool-call"}}} +{"type":"assistant/chunk","seq":10,"time":0,"data":{"turn":2,"step":1,"chunk":{"type":"tool-call-delta","index":0,"id":"delegate-write","name":"subagent","argumentsDelta":"{\"description\": \"Delegated write probe\", \"prompt\": \"Use the write tool exactly once with file_path set to exactly the relative path inherited.txt and content escaped. If the write is denied, reply with the single word CHILD_DENIED and the denial marker line; do not retry and do not request escalation. If it succeeds, reply CHILD_WROTE.\"}"}}} +{"type":"assistant/chunk","seq":11,"time":0,"data":{"turn":2,"step":1,"chunk":{"type":"block-end","index":0,"block":{"type":"tool-call","id":"delegate-write","name":"subagent","arguments":"{\"description\": \"Delegated write probe\", \"prompt\": \"Use the write tool exactly once with file_path set to exactly the relative path inherited.txt and content escaped. If the write is denied, reply with the single word CHILD_DENIED and the denial marker line; do not retry and do not request escalation. If it succeeds, reply CHILD_WROTE.\"}"}}}} +{"type":"assistant/chunk","seq":12,"time":0,"data":{"turn":2,"step":1,"chunk":{"type":"usage","usage":{"inputTokens":10,"outputTokens":5}}}} +{"type":"assistant/chunk","seq":13,"time":0,"data":{"turn":2,"step":1,"chunk":{"type":"finish","reason":{"kind":"tool-calls"}}}} +{"type":"assistant/message","seq":14,"time":0,"data":{"turn":2,"step":1,"message":{"role":"assistant","content":[{"type":"tool-call","id":"delegate-write","name":"subagent","arguments":"{\"description\": \"Delegated write probe\", \"prompt\": \"Use the write tool exactly once with file_path set to exactly the relative path inherited.txt and content escaped. If the write is denied, reply with the single word CHILD_DENIED and the denial marker line; do not retry and do not request escalation. If it succeeds, reply CHILD_WROTE.\"}"}],"source":{"kind":"model","provider":"deepseek","model":"deepseek-v4-flash"},"id":"{{sessionId}}"},"usage":{"inputTokens":10,"outputTokens":5}},"sourceEventSeqs":[9,10,11,12,13],"surfaceOp":"append"} +{"type":"tool/call","seq":15,"time":0,"data":{"turn":2,"step":1,"callId":"delegate-write","name":"subagent","arguments":"{\"description\": \"Delegated write probe\", \"prompt\": \"Use the write tool exactly once with file_path set to exactly the relative path inherited.txt and content escaped. If the write is denied, reply with the single word CHILD_DENIED and the denial marker line; do not retry and do not request escalation. If it succeeds, reply CHILD_WROTE.\"}"}} +{"type":"tool/result","seq":16,"time":0,"data":{"turn":2,"step":1,"message":{"source":{"kind":"tool","callId":"delegate-write"},"content":[{"type":"tool-result","toolCallId":"delegate-write","content":[{"type":"text","text":"CHILD_DENIED [sandbox: file access denied under read-only mode]"}],"isError":false}],"role":"user","id":"{{sessionId}}"}},"sourceEventSeqs":[15],"surfaceOp":"append"} +{"type":"step/end","seq":17,"time":0,"data":{"turn":2,"step":1}} +{"type":"step/start","seq":18,"time":0,"data":{"turn":2,"step":2}} +{"type":"assistant/chunk","seq":19,"time":0,"data":{"turn":2,"step":2,"chunk":{"type":"block-start","index":0,"blockType":"text"}}} +{"type":"assistant/chunk","seq":20,"time":0,"data":{"turn":2,"step":2,"chunk":{"type":"text-delta","index":0,"text":"The delegated child was denied by the sandbox. PARENT_DONE"}}} +{"type":"assistant/chunk","seq":21,"time":0,"data":{"turn":2,"step":2,"chunk":{"type":"block-end","index":0,"block":{"type":"text","text":"The delegated child was denied by the sandbox. PARENT_DONE"}}}} +{"type":"assistant/chunk","seq":22,"time":0,"data":{"turn":2,"step":2,"chunk":{"type":"usage","usage":{"inputTokens":10,"outputTokens":5}}}} +{"type":"assistant/chunk","seq":23,"time":0,"data":{"turn":2,"step":2,"chunk":{"type":"finish","reason":{"kind":"stop"}}}} +{"type":"assistant/message","seq":24,"time":0,"data":{"turn":2,"step":2,"message":{"role":"assistant","content":[{"type":"text","text":"The delegated child was denied by the sandbox. PARENT_DONE"}],"source":{"kind":"model","provider":"deepseek","model":"deepseek-v4-flash"},"id":"{{sessionId}}"},"usage":{"inputTokens":10,"outputTokens":5}},"sourceEventSeqs":[19,20,21,22,23],"surfaceOp":"append"} +{"type":"step/end","seq":25,"time":0,"data":{"turn":2,"step":2}} +{"type":"turn/end","seq":26,"time":0,"data":{"turn":2,"reason":{"kind":"completed"}}} diff --git a/examples/headless-agent/tests/subagent-inheritance-snapshots/parent-override/replay.override.json b/examples/headless-agent/tests/subagent-inheritance-snapshots/parent-override/replay.override.json new file mode 100644 index 0000000000..1c0fe9ffad --- /dev/null +++ b/examples/headless-agent/tests/subagent-inheritance-snapshots/parent-override/replay.override.json @@ -0,0 +1 @@ +[{"kind": "chunks", "chunks": [{"type": "block-start", "index": 0, "blockType": "tool-call"}, {"type": "tool-call-delta", "index": 0, "id": "delegate-write", "name": "subagent", "argumentsDelta": "{\"description\": \"Delegated write probe\", \"prompt\": \"Use the write tool exactly once with file_path set to exactly the relative path inherited.txt and content escaped. If the write is denied, reply with the single word CHILD_DENIED and the denial marker line; do not retry and do not request escalation. If it succeeds, reply CHILD_WROTE.\"}"}, {"type": "block-end", "index": 0, "block": {"type": "tool-call", "id": "delegate-write", "name": "subagent", "arguments": "{\"description\": \"Delegated write probe\", \"prompt\": \"Use the write tool exactly once with file_path set to exactly the relative path inherited.txt and content escaped. If the write is denied, reply with the single word CHILD_DENIED and the denial marker line; do not retry and do not request escalation. If it succeeds, reply CHILD_WROTE.\"}"}}, {"type": "usage", "usage": {"inputTokens": 10, "outputTokens": 5}}, {"type": "finish", "reason": {"kind": "tool-calls"}}]}, {"kind": "chunks", "chunks": [{"type": "block-start", "index": 0, "blockType": "text"}, {"type": "text-delta", "index": 0, "text": "The delegated child was denied by the sandbox. PARENT_DONE"}, {"type": "block-end", "index": 0, "block": {"type": "text", "text": "The delegated child was denied by the sandbox. PARENT_DONE"}}, {"type": "usage", "usage": {"inputTokens": 10, "outputTokens": 5}}, {"type": "finish", "reason": {"kind": "stop"}}]}] diff --git a/examples/headless-agent/tests/subagent-inheritance.snapshot.ts b/examples/headless-agent/tests/subagent-inheritance.snapshot.ts new file mode 100644 index 0000000000..02c96cac15 --- /dev/null +++ b/examples/headless-agent/tests/subagent-inheritance.snapshot.ts @@ -0,0 +1,124 @@ +/** + * Assembled-app regression: a parent-only read-only override is seeded into + * its child log and confines a real write under a wider deployment default. + */ + +import { readFile, readdir, writeFile } from 'node:fs/promises' +import { join } from 'node:path' +import { fileURLToPath } from 'node:url' +import { Context } from 'cordis' +import { normalizeSessionLog, scrubRequestHeaders, type NormalizeContext } from '@deepseek-ai/dsh-acp-snapshot' +import { LOADER_SMOKE_TEST_TIMEOUT_MS, runLoaderSmoke } from '@deepseek-ai/dsh-loader-smoke' +import { createUserMessage } from '@deepseek-ai/dsh-llm' +import SessionStore, { SESSION_FORMAT_VERSION, SessionId, type SessionEvent, type SessionHeader } from '@deepseek-ai/dsh-session' +import SessionPersistenceJsonl from '@deepseek-ai/dsh-session-persistence-jsonl' +import { describe, expect, it } from 'vitest' + +const fixtureDir = fileURLToPath(new URL('./subagent-inheritance-snapshots/parent-override', import.meta.url)) +const replayOverride = join(fixtureDir, 'replay.override.json') +const childReplay = join(fixtureDir, 'child.replay.jsonl') +const parentExpected = join(fixtureDir, 'parent.expected.jsonl') +const childExpected = join(fixtureDir, 'child.expected.jsonl') +const configPath = fileURLToPath(new URL('../subagent-inheritance.cordis.snapshot.yml', import.meta.url)) +const binScript = fileURLToPath(new URL('../../../packages/examples/cli-demo/src/bin.ts', import.meta.url)) +const tsconfigPath = fileURLToPath(new URL('../../../tsconfig.json', import.meta.url)) +const sessionId = SessionId('subagent-inheritance-parent') +const refreshing = process.env.DSH_SNAPSHOT === 'refresh' +const task = 'Delegate the write probe to a subagent.' + +/** Seed a completed parent turn with the only read-only fact in the app. */ +async function seedReadOnlyParent(root: string, cwd: string): Promise { + const ctx = new Context() + await ctx.plugin(SessionStore) + await ctx.plugin(SessionPersistenceJsonl, { root, compression: 'none' }) + const meta: SessionHeader = { + version: SESSION_FORMAT_VERSION, + id: sessionId, + createdAt: 1, + cwd, + delegationDepth: 0, + } + const events: SessionEvent[] = [ + { type: 'turn/start', seq: 0, time: 10, data: { turn: 1, trigger: { kind: 'message', source: { kind: 'user' } } } }, + { type: 'user/message', seq: 1, time: 11, data: createUserMessage({ content: [{ type: 'text', text: 'Tighten this session to read-only.' }], source: { kind: 'user' } }), surfaceOp: 'append' }, + { type: 'sandbox/mode', seq: 2, time: 12, data: { mode: 'read-only' } }, + { type: 'turn/end', seq: 3, time: 13, data: { turn: 1, reason: { kind: 'completed' } } }, + ] + try { + await ctx.sessionPersistence.create(meta) + await ctx.sessionPersistence.append(sessionId, events) + } finally { + await ctx.fiber.dispose() + } +} + +describe('parent-only override inheritance snapshot', () => { + it('confines a delegated child through the assembled headless app', async () => { + let cwd = '' + const result = await runLoaderSmoke({ + label: 'subagent inheritance headless stream-json snapshot', + tempDirPrefix: 'dsh-subagent-inherit-', + binScript, + configPath, + binArgs: ['--config', configPath, '--output-format', 'stream-json', task], + tsconfigPath, + env: { + // The primary fixture path must exist for llm-replay's config guard; + // the override sidecar fully replaces the derived parent script. + DSH_SNAPSHOT_FILE: replayOverride, + DSH_SNAPSHOT_OVERRIDE: replayOverride, + DSH_SNAPSHOT_CHILD_FILES: childReplay, + }, + prepare: async (runCwd) => { + cwd = runCwd + await seedReadOnlyParent(join(runCwd, '.sessions'), runCwd) + }, + inspect: async (runCwd) => { + // THE physical fact: the child's write never reached the disk. Under + // the deployment default (workspace-write) alone it would succeed. + await expect(readFile(join(runCwd, 'inherited.txt'), 'utf8')).rejects.toMatchObject({ code: 'ENOENT' }) + + // Collect both persisted logs (parent resumed turn + child run). + const sessionsDir = join(runCwd, '.sessions') + const files = (await readdir(sessionsDir, { recursive: true })).filter(file => file.endsWith('.jsonl')) + const logs = await Promise.all(files.map(async file => readFile(join(sessionsDir, file), 'utf8'))) + const headerOf = (content: string): Record => + JSON.parse(content.split('\n')[0] ?? '{}') as Record + const parent = logs.find(content => content.includes('"subagent-inheritance-parent"')) + const child = logs.find(content => typeof headerOf(content).parentSession === 'string') + if (parent === undefined || child === undefined) throw new Error('missing persisted parent or child log') + + const childRecords = child.trimEnd().split('\n').map( + line => JSON.parse(line) as Record, + ) + expect(childRecords[1]).toMatchObject({ + type: 'sandbox/mode', + seq: 0, + data: { mode: 'read-only', source: 'delegation' }, + }) + + const context: NormalizeContext = { sessionIds: [sessionId, String(headerOf(child).id)], cwd } + const normalizedParent = scrubRequestHeaders(normalizeSessionLog(parent, context)) + const normalizedChild = scrubRequestHeaders(normalizeSessionLog(child, context)) + if (refreshing) { + await writeFile(parentExpected, normalizedParent) + await writeFile(childExpected, normalizedChild) + } + expect(normalizedParent).toBe(await readFile(parentExpected, 'utf8')) + expect(normalizedChild).toBe(await readFile(childExpected, 'utf8')) + // The child's real write was denied by the real fence. + expect(normalizedChild).toContain('file access denied under read-only mode') + }, + }) + + expect(result.stderr).toBe('') + const records = result.stdout.trimEnd().split('\n').map(line => JSON.parse(line) as Record) + expect(records.at(-1)).toMatchObject({ + type: 'result', + success: true, + sessionId, + result: 'The delegated child was denied by the sandbox. PARENT_DONE', + reason: { kind: 'completed' }, + }) + }, LOADER_SMOKE_TEST_TIMEOUT_MS) +}) diff --git a/examples/jsonrpc-agent/tests/sdk.snapshot.ts b/examples/jsonrpc-agent/tests/sdk.snapshot.ts index 26254f0cf1..c54812e3e5 100644 --- a/examples/jsonrpc-agent/tests/sdk.snapshot.ts +++ b/examples/jsonrpc-agent/tests/sdk.snapshot.ts @@ -20,6 +20,7 @@ import { refreshFixtureReplacements, scrubRequestHeaders, stabilizeRefreshLog, + tokenizeSessionFixtureCwd, type HarvestedLog, type NormalizeContext, } from '@deepseek-ai/dsh-acp-snapshot' @@ -238,7 +239,7 @@ describe('TypeScript SDK snapshots over the jsonrpc runtime', () => { await Promise.all(ordered.map(async (log, index) => { const file = fixtureFiles(scenario)[index] if (file === undefined) throw new Error(`no fixture path for persisted log ${index}`) - await writeFile(file, scrubRequestHeaders(log.content)) + await writeFile(file, scrubRequestHeaders(tokenizeSessionFixtureCwd(log.content))) })) } @@ -257,12 +258,19 @@ describe('TypeScript SDK snapshots over the jsonrpc runtime', () => { const existing = expectedContents[index] const file = files[index] if (existing === undefined || file === undefined) throw new Error(`no fixture for persisted log ${index}`) - const stable = stabilizeRefreshLog(log.content, existing, replacements, actualContext) + const stable = scrubRequestHeaders(tokenizeSessionFixtureCwd( + stabilizeRefreshLog(log.content, existing, replacements, actualContext), + )) await writeFile(file, stable) return stable })) } + for (const [index, expected] of expectedContents.entries()) { + expect(scrubRequestHeaders(expected), `${scenario.name} session fixture ${index} carries request-header bulk`) + .toBe(expected) + } + // Persisted transcripts match the committed fixtures. const expectedContext = contextOfContents(expectedContents) for (const [index, log] of ordered.entries()) { diff --git a/examples/jsonrpc-agent/tests/snapshots/bash-tool/session.jsonl b/examples/jsonrpc-agent/tests/snapshots/bash-tool/session.jsonl index d3ee0a2f5f..a1e3925531 100644 --- a/examples/jsonrpc-agent/tests/snapshots/bash-tool/session.jsonl +++ b/examples/jsonrpc-agent/tests/snapshots/bash-tool/session.jsonl @@ -1,9 +1,9 @@ -{"type":"session","version":0,"id":"sdk-snapshot-bash","createdAt":1785097395899,"cwd":"/tmp/sdk-snapshot-bash-tool-ywbuab","delegationDepth":0} +{"type":"session","version":0,"id":"sdk-snapshot-bash","createdAt":1785097395899,"cwd":"{{cwd}}","delegationDepth":0} {"type":"turn/start","seq":0,"time":1785097395904,"data":{"turn":1,"trigger":{"kind":"message","source":{"kind":"user"}}}} {"type":"user/message","seq":1,"time":1785097395905,"data":{"content":[{"type":"text","text":"Run this exact command with your bash tool, then reply with its stdout only: echo dsh-sdk-proof-7391"}],"source":{"kind":"user"},"role":"user","id":"295507c3-4ba7-4695-a535-73e75046abb3"},"surfaceOp":"append"} {"type":"session/title","seq":2,"time":1785097395907,"data":{"title":"Run this exact command with","messageSeqs":[1],"source":{"kind":"fallback"}}} {"type":"step/start","seq":3,"time":1785097395908,"data":{"turn":1,"step":1}} -{"type":"request/header","seq":4,"time":1785097395909,"data":{"header":{"config":{"provider":"deepseek","model":"deepseek-v4-flash"},"system":"You are an AI agent powered by the DeepSeek Harness SDK.\n\nYou are a coding agent.\n\nUse the read tool — not shell commands like cat — to inspect text files. Results include line numbers. Use offset and limit to continue reading large files.\n\nUse the write tool to create files or completely replace file contents. Existing files are overwritten, so read an existing file first (the default fs-policy requires it) and prefer edit for targeted changes.\n\nUse the edit tool for targeted changes to existing UTF-8 text files. It replaces literal old_string with new_string; by default old_string must appear exactly once. If old_string appears multiple times, provide a more specific old_string or set replace_all to true. Read the file first (the default fs-policy requires it), unless you just created or edited it in this session.\n\nCheck the [exit code: N] marker on every bash result; investigate failures before moving on.","tools":[{"name":"bash","description":"Execute a bash command (`bash -c`) and return its stdout/stderr. Each call runs in a fresh shell: no state (cwd, variables, functions) persists between calls — pass `workdir` instead of using `cd`. Non-zero exits are reported as `[exit code: N]`. Current harness environment facts are exposed through managed `$DSH_*` variables; inspect them when needed. Commands may run under a file sandbox; a blocked file operation is reported as `[sandbox: file access denied under mode]` — a policy denial, not a bug in the command; do not retry another way. Long output is truncated to its tail; the full output is saved to a file whose path is reported when available. Background execution is not available; long-running commands must finish within the timeout.","parameters":{"type":"object","properties":{"command":{"type":"string","description":"The bash command to execute."},"description":{"type":"string","description":"Clear, concise description of what this command does in active voice, 5-10 words (shown in the UI). Examples: \"ls\" → \"List files in current directory\"; \"git status\" → \"Show working tree status\"; \"npm install\" → \"Install package dependencies\"."},"timeoutMs":{"type":"number","description":"Timeout in milliseconds. The executor applies its configured default and cap, and kills the command on expiry."},"workdir":{"type":"string","description":"Working directory for this command. Defaults to the session workspace; a relative path is resolved against it."}},"required":["command","description"]}},{"name":"edit","description":"Edit an existing UTF-8 text file by replacing literal text.","parameters":{"type":"object","properties":{"file_path":{"type":"string","description":"Path to edit, resolved by the filesystem backend."},"old_string":{"type":"string","description":"Literal text to replace. Must match exactly."},"new_string":{"type":"string","description":"Literal replacement text. Use an empty string to delete the match."},"replace_all":{"type":"boolean","description":"Replace all matches. Defaults to false; when false, old_string must appear exactly once."}},"required":["file_path","old_string","new_string"]}},{"name":"read","description":"Read a UTF-8 text file and return line-numbered content.","parameters":{"type":"object","properties":{"file_path":{"type":"string","description":"Path to read, resolved by the filesystem backend."},"offset":{"type":"number","description":"1-based first line to return. Defaults to 1."},"limit":{"type":"number","description":"Maximum number of lines to return. Defaults to 2000."}},"required":["file_path"]}},{"name":"subagent","description":"Delegate a self-contained task to a subagent (a separate agent that works in its own context) and return its final result. Use this to offload focused, independent work — research, a scoped implementation, an analysis — so it does not consume this conversation's context. The subagent runs to completion and you receive only its final answer, not its intermediate steps. Give it a complete, standalone prompt: it does not see this conversation.","parameters":{"type":"object","properties":{"description":{"type":"string","description":"A short (3-5 word) description of the delegated task, for display."},"prompt":{"type":"string","description":"The complete, self-contained task for the subagent. It does not share this conversation's context, so include everything it needs."}},"required":["description","prompt"]}},{"name":"todo_write","description":"Record and update a structured task list for the current work. Send the ENTIRE list every call — it REPLACES the previous list (there are no partial updates, no per-item edits). Use it to plan multi-step work and show progress: add one todo per concrete step before you start. Keep AT MOST ONE todo `in_progress` at a time; while work remains, exactly one active task should be `in_progress`. Mark a todo `completed` the moment it is done (do not batch completions), and allow no `in_progress` item only once all work is complete. Skip the list for trivial single-step tasks. Statuses: `pending` (not started), `in_progress` (being worked on now), `completed` (finished).","parameters":{"type":"object","properties":{"todos":{"type":"array","description":"The COMPLETE task list, replacing any previous list.","items":{"type":"object","additionalProperties":false,"properties":{"content":{"type":"string","description":"What the task is — a short imperative line."},"status":{"type":"string","description":"pending (not started) | in_progress (now) | completed (done).","enum":["pending","in_progress","completed"]}},"required":["content","status"]}}},"required":["todos"]}},{"name":"write","description":"Create or fully replace a UTF-8 text file.","parameters":{"type":"object","properties":{"file_path":{"type":"string","description":"Path to write, resolved by the filesystem backend."},"content":{"type":"string","description":"Full UTF-8 text content to write."}},"required":["file_path","content"]}}]},"reason":"initial"}} +{"type":"request/header","seq":4,"time":1785097395909,"data":{"header":{"config":{"provider":"deepseek","model":"deepseek-v4-flash"},"system":"{{system}}","tools":"{{tools}}"},"reason":"initial"}} {"type":"assistant/chunk","seq":5,"time":1785097396437,"data":{"turn":1,"step":1,"chunk":{"type":"block-start","index":0,"blockType":"reasoning"}}} {"type":"reasoning-chunks","seq0":6,"time0":1785097396438,"data":{"turn":1,"step":1,"index":0,"dt":[219,22,1,0,0,0,1,24,25,0,0,25,1,24,1,0],"texts":["The"," user"," wants"," me"," to"," run"," a"," specific"," bash"," command"," and"," reply"," with"," its"," stdout"," only","."]}} {"type":"assistant/chunk","seq":23,"time":1785097396856,"data":{"turn":1,"step":1,"chunk":{"type":"block-start","index":1,"blockType":"tool-call"}}} diff --git a/examples/jsonrpc-agent/tests/snapshots/subagent-spawn/session.1.jsonl b/examples/jsonrpc-agent/tests/snapshots/subagent-spawn/session.1.jsonl index 0e9fb1c561..7b0db305f6 100644 --- a/examples/jsonrpc-agent/tests/snapshots/subagent-spawn/session.1.jsonl +++ b/examples/jsonrpc-agent/tests/snapshots/subagent-spawn/session.1.jsonl @@ -1,9 +1,9 @@ -{"type":"session","version":0,"id":"0b7fd85c-9f6f-4d46-b954-363984ce66fb","createdAt":1785097410282,"cwd":"/tmp/sdk-snapshot-subagent-spawn-6fzuBd","parentSession":"sdk-snapshot-subagent","delegationDepth":1} +{"type":"session","version":0,"id":"0b7fd85c-9f6f-4d46-b954-363984ce66fb","createdAt":1785097410282,"cwd":"{{cwd}}","parentSession":"sdk-snapshot-subagent","delegationDepth":1} {"type":"turn/start","seq":0,"time":1785097410283,"data":{"turn":1,"trigger":{"kind":"message","source":{"kind":"user"}}}} {"type":"user/message","seq":1,"time":1785097410283,"data":{"content":[{"type":"text","text":"Reply with exactly: child answer 42."}],"source":{"kind":"user"},"role":"user","id":"fb1dfb09-5b8b-4343-8a04-49cc4c7c082e"},"surfaceOp":"append"} {"type":"session/title","seq":2,"time":1785097410283,"data":{"title":"Reply with exactly: child answer","messageSeqs":[1],"source":{"kind":"fallback"}}} {"type":"step/start","seq":3,"time":1785097410284,"data":{"turn":1,"step":1}} -{"type":"request/header","seq":4,"time":1785097410284,"data":{"header":{"config":{"provider":"deepseek","model":"deepseek-v4-flash"},"system":"You are an AI agent powered by the DeepSeek Harness SDK.\n\nYou are a coding agent.\n\nUse the read tool — not shell commands like cat — to inspect text files. Results include line numbers. Use offset and limit to continue reading large files.\n\nUse the write tool to create files or completely replace file contents. Existing files are overwritten, so read an existing file first (the default fs-policy requires it) and prefer edit for targeted changes.\n\nUse the edit tool for targeted changes to existing UTF-8 text files. It replaces literal old_string with new_string; by default old_string must appear exactly once. If old_string appears multiple times, provide a more specific old_string or set replace_all to true. Read the file first (the default fs-policy requires it), unless you just created or edited it in this session.\n\nCheck the [exit code: N] marker on every bash result; investigate failures before moving on.","tools":[{"name":"bash","description":"Execute a bash command (`bash -c`) and return its stdout/stderr. Each call runs in a fresh shell: no state (cwd, variables, functions) persists between calls — pass `workdir` instead of using `cd`. Non-zero exits are reported as `[exit code: N]`. Current harness environment facts are exposed through managed `$DSH_*` variables; inspect them when needed. Commands may run under a file sandbox; a blocked file operation is reported as `[sandbox: file access denied under mode]` — a policy denial, not a bug in the command; do not retry another way. Long output is truncated to its tail; the full output is saved to a file whose path is reported when available. Background execution is not available; long-running commands must finish within the timeout.","parameters":{"type":"object","properties":{"command":{"type":"string","description":"The bash command to execute."},"description":{"type":"string","description":"Clear, concise description of what this command does in active voice, 5-10 words (shown in the UI). Examples: \"ls\" → \"List files in current directory\"; \"git status\" → \"Show working tree status\"; \"npm install\" → \"Install package dependencies\"."},"timeoutMs":{"type":"number","description":"Timeout in milliseconds. The executor applies its configured default and cap, and kills the command on expiry."},"workdir":{"type":"string","description":"Working directory for this command. Defaults to the session workspace; a relative path is resolved against it."}},"required":["command","description"]}},{"name":"edit","description":"Edit an existing UTF-8 text file by replacing literal text.","parameters":{"type":"object","properties":{"file_path":{"type":"string","description":"Path to edit, resolved by the filesystem backend."},"old_string":{"type":"string","description":"Literal text to replace. Must match exactly."},"new_string":{"type":"string","description":"Literal replacement text. Use an empty string to delete the match."},"replace_all":{"type":"boolean","description":"Replace all matches. Defaults to false; when false, old_string must appear exactly once."}},"required":["file_path","old_string","new_string"]}},{"name":"read","description":"Read a UTF-8 text file and return line-numbered content.","parameters":{"type":"object","properties":{"file_path":{"type":"string","description":"Path to read, resolved by the filesystem backend."},"offset":{"type":"number","description":"1-based first line to return. Defaults to 1."},"limit":{"type":"number","description":"Maximum number of lines to return. Defaults to 2000."}},"required":["file_path"]}},{"name":"subagent","description":"Delegate a self-contained task to a subagent (a separate agent that works in its own context) and return its final result. Use this to offload focused, independent work — research, a scoped implementation, an analysis — so it does not consume this conversation's context. The subagent runs to completion and you receive only its final answer, not its intermediate steps. Give it a complete, standalone prompt: it does not see this conversation.","parameters":{"type":"object","properties":{"description":{"type":"string","description":"A short (3-5 word) description of the delegated task, for display."},"prompt":{"type":"string","description":"The complete, self-contained task for the subagent. It does not share this conversation's context, so include everything it needs."}},"required":["description","prompt"]}},{"name":"todo_write","description":"Record and update a structured task list for the current work. Send the ENTIRE list every call — it REPLACES the previous list (there are no partial updates, no per-item edits). Use it to plan multi-step work and show progress: add one todo per concrete step before you start. Keep AT MOST ONE todo `in_progress` at a time; while work remains, exactly one active task should be `in_progress`. Mark a todo `completed` the moment it is done (do not batch completions), and allow no `in_progress` item only once all work is complete. Skip the list for trivial single-step tasks. Statuses: `pending` (not started), `in_progress` (being worked on now), `completed` (finished).","parameters":{"type":"object","properties":{"todos":{"type":"array","description":"The COMPLETE task list, replacing any previous list.","items":{"type":"object","additionalProperties":false,"properties":{"content":{"type":"string","description":"What the task is — a short imperative line."},"status":{"type":"string","description":"pending (not started) | in_progress (now) | completed (done).","enum":["pending","in_progress","completed"]}},"required":["content","status"]}}},"required":["todos"]}},{"name":"write","description":"Create or fully replace a UTF-8 text file.","parameters":{"type":"object","properties":{"file_path":{"type":"string","description":"Path to write, resolved by the filesystem backend."},"content":{"type":"string","description":"Full UTF-8 text content to write."}},"required":["file_path","content"]}}]},"reason":"initial"}} +{"type":"request/header","seq":4,"time":1785097410284,"data":{"header":{"config":{"provider":"deepseek","model":"deepseek-v4-flash"},"system":"{{system}}","tools":"{{tools}}"},"reason":"initial"}} {"type":"assistant/chunk","seq":5,"time":1785097410836,"data":{"turn":1,"step":1,"chunk":{"type":"block-start","index":0,"blockType":"reasoning"}}} {"type":"reasoning-chunks","seq0":6,"time0":1785097410836,"data":{"turn":1,"step":1,"index":0,"dt":[149,26,0,0,24,1,0,0,0,25,0,1,0],"texts":["The"," user"," wants"," me"," to"," reply"," with"," exactly"," \"","child"," answer"," ","42",".\""]}} {"type":"assistant/chunk","seq":20,"time":1785097411113,"data":{"turn":1,"step":1,"chunk":{"type":"block-start","index":1,"blockType":"text"}}} diff --git a/examples/jsonrpc-agent/tests/snapshots/subagent-spawn/session.jsonl b/examples/jsonrpc-agent/tests/snapshots/subagent-spawn/session.jsonl index 23127d77b8..f43a78f588 100644 --- a/examples/jsonrpc-agent/tests/snapshots/subagent-spawn/session.jsonl +++ b/examples/jsonrpc-agent/tests/snapshots/subagent-spawn/session.jsonl @@ -1,9 +1,9 @@ -{"type":"session","version":0,"id":"sdk-snapshot-subagent","createdAt":1785097408901,"cwd":"/tmp/sdk-snapshot-subagent-spawn-6fzuBd","delegationDepth":0} +{"type":"session","version":0,"id":"sdk-snapshot-subagent","createdAt":1785097408901,"cwd":"{{cwd}}","delegationDepth":0} {"type":"turn/start","seq":0,"time":1785097408905,"data":{"turn":1,"trigger":{"kind":"message","source":{"kind":"user"}}}} {"type":"user/message","seq":1,"time":1785097408905,"data":{"content":[{"type":"text","text":"Use the subagent tool exactly once with description 'echo probe' and prompt: Reply with exactly: child answer 42. Then reply with the subagent's final answer verbatim."}],"source":{"kind":"user"},"role":"user","id":"e2664740-19d2-4e54-81e5-63ff154af28e"},"surfaceOp":"append"} {"type":"session/title","seq":2,"time":1785097408907,"data":{"title":"Use the subagent tool exactly","messageSeqs":[1],"source":{"kind":"fallback"}}} {"type":"step/start","seq":3,"time":1785097408908,"data":{"turn":1,"step":1}} -{"type":"request/header","seq":4,"time":1785097408908,"data":{"header":{"config":{"provider":"deepseek","model":"deepseek-v4-flash"},"system":"You are an AI agent powered by the DeepSeek Harness SDK.\n\nYou are a coding agent.\n\nUse the read tool — not shell commands like cat — to inspect text files. Results include line numbers. Use offset and limit to continue reading large files.\n\nUse the write tool to create files or completely replace file contents. Existing files are overwritten, so read an existing file first (the default fs-policy requires it) and prefer edit for targeted changes.\n\nUse the edit tool for targeted changes to existing UTF-8 text files. It replaces literal old_string with new_string; by default old_string must appear exactly once. If old_string appears multiple times, provide a more specific old_string or set replace_all to true. Read the file first (the default fs-policy requires it), unless you just created or edited it in this session.\n\nCheck the [exit code: N] marker on every bash result; investigate failures before moving on.","tools":[{"name":"bash","description":"Execute a bash command (`bash -c`) and return its stdout/stderr. Each call runs in a fresh shell: no state (cwd, variables, functions) persists between calls — pass `workdir` instead of using `cd`. Non-zero exits are reported as `[exit code: N]`. Current harness environment facts are exposed through managed `$DSH_*` variables; inspect them when needed. Commands may run under a file sandbox; a blocked file operation is reported as `[sandbox: file access denied under mode]` — a policy denial, not a bug in the command; do not retry another way. Long output is truncated to its tail; the full output is saved to a file whose path is reported when available. Background execution is not available; long-running commands must finish within the timeout.","parameters":{"type":"object","properties":{"command":{"type":"string","description":"The bash command to execute."},"description":{"type":"string","description":"Clear, concise description of what this command does in active voice, 5-10 words (shown in the UI). Examples: \"ls\" → \"List files in current directory\"; \"git status\" → \"Show working tree status\"; \"npm install\" → \"Install package dependencies\"."},"timeoutMs":{"type":"number","description":"Timeout in milliseconds. The executor applies its configured default and cap, and kills the command on expiry."},"workdir":{"type":"string","description":"Working directory for this command. Defaults to the session workspace; a relative path is resolved against it."}},"required":["command","description"]}},{"name":"edit","description":"Edit an existing UTF-8 text file by replacing literal text.","parameters":{"type":"object","properties":{"file_path":{"type":"string","description":"Path to edit, resolved by the filesystem backend."},"old_string":{"type":"string","description":"Literal text to replace. Must match exactly."},"new_string":{"type":"string","description":"Literal replacement text. Use an empty string to delete the match."},"replace_all":{"type":"boolean","description":"Replace all matches. Defaults to false; when false, old_string must appear exactly once."}},"required":["file_path","old_string","new_string"]}},{"name":"read","description":"Read a UTF-8 text file and return line-numbered content.","parameters":{"type":"object","properties":{"file_path":{"type":"string","description":"Path to read, resolved by the filesystem backend."},"offset":{"type":"number","description":"1-based first line to return. Defaults to 1."},"limit":{"type":"number","description":"Maximum number of lines to return. Defaults to 2000."}},"required":["file_path"]}},{"name":"subagent","description":"Delegate a self-contained task to a subagent (a separate agent that works in its own context) and return its final result. Use this to offload focused, independent work — research, a scoped implementation, an analysis — so it does not consume this conversation's context. The subagent runs to completion and you receive only its final answer, not its intermediate steps. Give it a complete, standalone prompt: it does not see this conversation.","parameters":{"type":"object","properties":{"description":{"type":"string","description":"A short (3-5 word) description of the delegated task, for display."},"prompt":{"type":"string","description":"The complete, self-contained task for the subagent. It does not share this conversation's context, so include everything it needs."}},"required":["description","prompt"]}},{"name":"todo_write","description":"Record and update a structured task list for the current work. Send the ENTIRE list every call — it REPLACES the previous list (there are no partial updates, no per-item edits). Use it to plan multi-step work and show progress: add one todo per concrete step before you start. Keep AT MOST ONE todo `in_progress` at a time; while work remains, exactly one active task should be `in_progress`. Mark a todo `completed` the moment it is done (do not batch completions), and allow no `in_progress` item only once all work is complete. Skip the list for trivial single-step tasks. Statuses: `pending` (not started), `in_progress` (being worked on now), `completed` (finished).","parameters":{"type":"object","properties":{"todos":{"type":"array","description":"The COMPLETE task list, replacing any previous list.","items":{"type":"object","additionalProperties":false,"properties":{"content":{"type":"string","description":"What the task is — a short imperative line."},"status":{"type":"string","description":"pending (not started) | in_progress (now) | completed (done).","enum":["pending","in_progress","completed"]}},"required":["content","status"]}}},"required":["todos"]}},{"name":"write","description":"Create or fully replace a UTF-8 text file.","parameters":{"type":"object","properties":{"file_path":{"type":"string","description":"Path to write, resolved by the filesystem backend."},"content":{"type":"string","description":"Full UTF-8 text content to write."}},"required":["file_path","content"]}}]},"reason":"initial"}} +{"type":"request/header","seq":4,"time":1785097408908,"data":{"header":{"config":{"provider":"deepseek","model":"deepseek-v4-flash"},"system":"{{system}}","tools":"{{tools}}"},"reason":"initial"}} {"type":"assistant/chunk","seq":5,"time":1785097409495,"data":{"turn":1,"step":1,"chunk":{"type":"block-start","index":0,"blockType":"reasoning"}}} {"type":"reasoning-chunks","seq0":6,"time0":1785097409496,"data":{"turn":1,"step":1,"index":0,"dt":[170,25,1,0,0,0,0,24,0,1,0,0,0,26,0,0,0,0,0,26,0,0,0,0,0,30,0,0,1,0,0,20,1,0,0,28,0,1,0,0,0,23,1,0,0,0,0,25,1,25,26,1,0,0],"texts":["The"," user"," wants"," me"," to",":\n","1","."," Use"," the"," sub","agent"," tool"," exactly"," once"," with"," description"," '","echo"," probe","'"," and"," prompt"," '","Reply"," with"," exactly",":"," child"," answer"," ","42",".'\n","2","."," Then"," reply"," with"," the"," sub","agent","'s"," final"," answer"," verb","atim",".\n\n","Let"," me"," do"," this"," step"," by"," step","."]}} {"type":"assistant/chunk","seq":61,"time":1785097410031,"data":{"turn":1,"step":1,"chunk":{"type":"block-start","index":1,"blockType":"tool-call"}}} diff --git a/examples/jsonrpc-agent/tests/snapshots/text-turn/session.jsonl b/examples/jsonrpc-agent/tests/snapshots/text-turn/session.jsonl index f3706e24d0..d7192b0a12 100644 --- a/examples/jsonrpc-agent/tests/snapshots/text-turn/session.jsonl +++ b/examples/jsonrpc-agent/tests/snapshots/text-turn/session.jsonl @@ -1,9 +1,9 @@ -{"type":"session","version":0,"id":"sdk-snapshot-text","createdAt":1785097381464,"cwd":"/tmp/sdk-snapshot-text-turn-OwFEJv","delegationDepth":0} +{"type":"session","version":0,"id":"sdk-snapshot-text","createdAt":1785097381464,"cwd":"{{cwd}}","delegationDepth":0} {"type":"turn/start","seq":0,"time":1785097381468,"data":{"turn":1,"trigger":{"kind":"message","source":{"kind":"user"}}}} {"type":"user/message","seq":1,"time":1785097381469,"data":{"content":[{"type":"text","text":"Reply with exactly: SDK snapshot OK"}],"source":{"kind":"user"},"role":"user","id":"4cb523e7-19c9-45d0-8799-911a78c26207"},"surfaceOp":"append"} {"type":"session/title","seq":2,"time":1785097381471,"data":{"title":"Reply with exactly: SDK snapshot","messageSeqs":[1],"source":{"kind":"fallback"}}} {"type":"step/start","seq":3,"time":1785097381472,"data":{"turn":1,"step":1}} -{"type":"request/header","seq":4,"time":1785097381472,"data":{"header":{"config":{"provider":"deepseek","model":"deepseek-v4-flash"},"system":"You are an AI agent powered by the DeepSeek Harness SDK.\n\nYou are a coding agent.\n\nUse the read tool — not shell commands like cat — to inspect text files. Results include line numbers. Use offset and limit to continue reading large files.\n\nUse the write tool to create files or completely replace file contents. Existing files are overwritten, so read an existing file first (the default fs-policy requires it) and prefer edit for targeted changes.\n\nUse the edit tool for targeted changes to existing UTF-8 text files. It replaces literal old_string with new_string; by default old_string must appear exactly once. If old_string appears multiple times, provide a more specific old_string or set replace_all to true. Read the file first (the default fs-policy requires it), unless you just created or edited it in this session.\n\nCheck the [exit code: N] marker on every bash result; investigate failures before moving on.","tools":[{"name":"bash","description":"Execute a bash command (`bash -c`) and return its stdout/stderr. Each call runs in a fresh shell: no state (cwd, variables, functions) persists between calls — pass `workdir` instead of using `cd`. Non-zero exits are reported as `[exit code: N]`. Current harness environment facts are exposed through managed `$DSH_*` variables; inspect them when needed. Commands may run under a file sandbox; a blocked file operation is reported as `[sandbox: file access denied under mode]` — a policy denial, not a bug in the command; do not retry another way. Long output is truncated to its tail; the full output is saved to a file whose path is reported when available. Background execution is not available; long-running commands must finish within the timeout.","parameters":{"type":"object","properties":{"command":{"type":"string","description":"The bash command to execute."},"description":{"type":"string","description":"Clear, concise description of what this command does in active voice, 5-10 words (shown in the UI). Examples: \"ls\" → \"List files in current directory\"; \"git status\" → \"Show working tree status\"; \"npm install\" → \"Install package dependencies\"."},"timeoutMs":{"type":"number","description":"Timeout in milliseconds. The executor applies its configured default and cap, and kills the command on expiry."},"workdir":{"type":"string","description":"Working directory for this command. Defaults to the session workspace; a relative path is resolved against it."}},"required":["command","description"]}},{"name":"edit","description":"Edit an existing UTF-8 text file by replacing literal text.","parameters":{"type":"object","properties":{"file_path":{"type":"string","description":"Path to edit, resolved by the filesystem backend."},"old_string":{"type":"string","description":"Literal text to replace. Must match exactly."},"new_string":{"type":"string","description":"Literal replacement text. Use an empty string to delete the match."},"replace_all":{"type":"boolean","description":"Replace all matches. Defaults to false; when false, old_string must appear exactly once."}},"required":["file_path","old_string","new_string"]}},{"name":"read","description":"Read a UTF-8 text file and return line-numbered content.","parameters":{"type":"object","properties":{"file_path":{"type":"string","description":"Path to read, resolved by the filesystem backend."},"offset":{"type":"number","description":"1-based first line to return. Defaults to 1."},"limit":{"type":"number","description":"Maximum number of lines to return. Defaults to 2000."}},"required":["file_path"]}},{"name":"subagent","description":"Delegate a self-contained task to a subagent (a separate agent that works in its own context) and return its final result. Use this to offload focused, independent work — research, a scoped implementation, an analysis — so it does not consume this conversation's context. The subagent runs to completion and you receive only its final answer, not its intermediate steps. Give it a complete, standalone prompt: it does not see this conversation.","parameters":{"type":"object","properties":{"description":{"type":"string","description":"A short (3-5 word) description of the delegated task, for display."},"prompt":{"type":"string","description":"The complete, self-contained task for the subagent. It does not share this conversation's context, so include everything it needs."}},"required":["description","prompt"]}},{"name":"todo_write","description":"Record and update a structured task list for the current work. Send the ENTIRE list every call — it REPLACES the previous list (there are no partial updates, no per-item edits). Use it to plan multi-step work and show progress: add one todo per concrete step before you start. Keep AT MOST ONE todo `in_progress` at a time; while work remains, exactly one active task should be `in_progress`. Mark a todo `completed` the moment it is done (do not batch completions), and allow no `in_progress` item only once all work is complete. Skip the list for trivial single-step tasks. Statuses: `pending` (not started), `in_progress` (being worked on now), `completed` (finished).","parameters":{"type":"object","properties":{"todos":{"type":"array","description":"The COMPLETE task list, replacing any previous list.","items":{"type":"object","additionalProperties":false,"properties":{"content":{"type":"string","description":"What the task is — a short imperative line."},"status":{"type":"string","description":"pending (not started) | in_progress (now) | completed (done).","enum":["pending","in_progress","completed"]}},"required":["content","status"]}}},"required":["todos"]}},{"name":"write","description":"Create or fully replace a UTF-8 text file.","parameters":{"type":"object","properties":{"file_path":{"type":"string","description":"Path to write, resolved by the filesystem backend."},"content":{"type":"string","description":"Full UTF-8 text content to write."}},"required":["file_path","content"]}}]},"reason":"initial"}} +{"type":"request/header","seq":4,"time":1785097381472,"data":{"header":{"config":{"provider":"deepseek","model":"deepseek-v4-flash"},"system":"{{system}}","tools":"{{tools}}"},"reason":"initial"}} {"type":"assistant/chunk","seq":5,"time":1785097381978,"data":{"turn":1,"step":1,"chunk":{"type":"block-start","index":0,"blockType":"reasoning"}}} {"type":"reasoning-chunks","seq0":6,"time0":1785097381979,"data":{"turn":1,"step":1,"index":0,"dt":[138,28,27,1,0,0,24,1,0,0,0,26,0,1,25,1,0,0],"texts":["The"," user"," wants"," me"," to"," reply"," with"," exactly"," \"","SD","K"," snapshot"," OK","\"."," Let"," me"," do"," that","."]}} {"type":"assistant/chunk","seq":25,"time":1785097382251,"data":{"turn":1,"step":1,"chunk":{"type":"block-start","index":1,"blockType":"text"}}} diff --git a/examples/tui-agent/tests/snapshots/bash-terminal-card/session.jsonl b/examples/tui-agent/tests/snapshots/bash-terminal-card/session.jsonl index f32d1160db..4d0354a167 100644 --- a/examples/tui-agent/tests/snapshots/bash-terminal-card/session.jsonl +++ b/examples/tui-agent/tests/snapshots/bash-terminal-card/session.jsonl @@ -1,4 +1,4 @@ -{"type":"session","version":0,"id":"e128dda9-ed11-4868-8266-0ef90d03c3d6","createdAt":1783352050748,"cwd":"/tmp/acp-snap-cwd-mrFUuk","delegationDepth":0} +{"type":"session","version":0,"id":"e128dda9-ed11-4868-8266-0ef90d03c3d6","createdAt":1783352050748,"cwd":"{{cwd}}","delegationDepth":0} {"type":"turn/start","seq":0,"time":1783352050753,"data":{"turn":1,"trigger":{"kind":"message","source":{"kind":"user"}}}} {"type":"user/message","seq":1,"time":1783352050753,"data":{"content":[{"type":"text","text":"Use the bash tool to run exactly: echo TERMINAL_OK. Then reply with the single word DONE and stop."}],"source":{"kind":"user"}},"surfaceOp":"append"} {"type":"step/start","seq":2,"time":1783352050755,"data":{"turn":1,"step":1}} diff --git a/examples/tui-agent/tests/snapshots/code-mode-dispatch-spill/session.jsonl b/examples/tui-agent/tests/snapshots/code-mode-dispatch-spill/session.jsonl index 42a7dcd8ea..3ae5c51857 100644 --- a/examples/tui-agent/tests/snapshots/code-mode-dispatch-spill/session.jsonl +++ b/examples/tui-agent/tests/snapshots/code-mode-dispatch-spill/session.jsonl @@ -1,4 +1,4 @@ -{"type":"session","version":0,"id":"main-session","createdAt":1785052797743,"cwd":"/tmp/dsh-tui-snapshot-code-mode-dispatch-spill-8cOdia"} +{"type":"session","version":0,"id":"main-session","createdAt":1785052797743,"cwd":"{{cwd}}"} {"type":"turn/start","seq":0,"time":1785052797817,"data":{"turn":1,"trigger":{"kind":"message","source":{"kind":"user"}}}} {"type":"user/message","seq":1,"time":1785052797818,"data":{"content":[{"type":"text","text":"Using ONE run_code program: call the bash tool exactly once with the command `seq 1 200 | awk '{printf \"line %04d: the quick brown fox jumps over the lazy dog\\n\", $1}'`, then return ONLY the number of lines in its output. Reply with just that number and stop."}],"source":{"kind":"user"}},"surfaceOp":"append"} {"type":"session/title","seq":2,"time":1785052797825,"data":{"title":"Using ONE run_code program: call","messageSeqs":[1],"source":{"kind":"fallback"}}} @@ -15,7 +15,7 @@ {"type":"assistant/message","seq":156,"time":1785052799798,"data":{"turn":1,"step":1,"content":[{"type":"reasoning","text":"The user wants me to write a single run_code program that calls bash exactly once with a specific command, then returns only the number of lines in its output."},{"type":"tool-call","id":"call_00_R6g9Uzx4h0jeUv9g3fno7490","name":"run_code","arguments":"{\"description\": \"Count lines in seq/awk output\", \"code\": \"const result = await tools.bash({\\n command: \\\"seq 1 200 | awk '{printf \\\\\\\"line %04d: the quick brown fox jumps over the lazy dog\\\\\\\\n\\\\\\\", $1}'\\\",\\n description: \\\"Generate 200 lines of text\\\"\\n});\\n\\n// Count lines in stdout\\nconst lines = result.kind === \\\"foreground\\\" ? result.stdout.text.trim().split(\\\"\\\\n\\\").length : 0;\\nreturn lines;\"}"}],"provenance":{"provider":"deepseek","model":"deepseek-v4-flash"},"usage":{"inputTokens":90,"outputTokens":186,"cacheReadTokens":3968,"reasoningTokens":32}},"sourceEventSeqs":[5,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20,21,22,23,24,25,26,27,28,29,30,31,32,33,34,35,36,37,38,39,40,41,42,43,44,45,46,47,48,49,50,51,52,53,54,55,56,57,58,59,60,61,62,63,64,65,66,67,68,69,70,71,72,73,74,75,76,77,78,79,80,81,82,83,84,85,86,87,88,89,90,91,92,93,94,95,96,97,98,99,100,101,102,103,104,105,106,107,108,109,110,111,112,113,114,115,116,117,118,119,120,121,122,123,124,125,126,127,128,129,130,131,132,133,134,135,136,137,138,139,140,141,142,143,144,145,146,147,148,149,150,151,152,153,154,155],"surfaceOp":"append"} {"type":"tool/call","seq":157,"time":1785052799799,"data":{"turn":1,"step":1,"callId":"call_00_R6g9Uzx4h0jeUv9g3fno7490","name":"run_code","arguments":"{\"description\": \"Count lines in seq/awk output\", \"code\": \"const result = await tools.bash({\\n command: \\\"seq 1 200 | awk '{printf \\\\\\\"line %04d: the quick brown fox jumps over the lazy dog\\\\\\\\n\\\\\\\", $1}'\\\",\\n description: \\\"Generate 200 lines of text\\\"\\n});\\n\\n// Count lines in stdout\\nconst lines = result.kind === \\\"foreground\\\" ? result.stdout.text.trim().split(\\\"\\\\n\\\").length : 0;\\nreturn lines;\"}"}} {"type":"tool/code-dispatch-start","seq":158,"time":1785052799893,"data":{"parentCallId":"call_00_R6g9Uzx4h0jeUv9g3fno7490","subCallId":"call_00_R6g9Uzx4h0jeUv9g3fno7490:code:1","name":"bash","arguments":{"command":"seq 1 200 | awk '{printf \"line %04d: the quick brown fox jumps over the lazy dog\\n\", $1}'","description":"Generate 200 lines of text"}}} -{"type":"tool/code-dispatch","seq":159,"time":1785052799923,"data":{"parentCallId":"call_00_R6g9Uzx4h0jeUv9g3fno7490","subCallId":"call_00_R6g9Uzx4h0jeUv9g3fno7490:code:1","name":"bash","arguments":{"command":"seq 1 200 | awk '{printf \"line %04d: the quick brown fox jumps over the lazy dog\\n\", $1}'","description":"Generate 200 lines of text"},"isError":false,"content":[{"type":"text","text":"line 0001: the quick brown fox jumps over the lazy dog\nline 0002: the quick brown fox jumps over the lazy dog\nline 0003: the quick brown fox jumps over the lazy dog\nline 0004: the quick s over the lazy dog\nline 0198: the quick brown fox jumps over the lazy dog\nline 0199: the quick brown fox jumps over the lazy dog\nline 0200: the quick brown fox jumps over the lazy dog\n\n\n(Omitted 10629 bytes. Full formatted result stored at: /tmp/dsh-tui-snapshot-code-mode-dispatch-spill-8cOdia/.spill/session-2d2b9e84a250/825a63550249-bash.txt. Use read with offset/limit, or grep this path to search within it.)"}]}} +{"type":"tool/code-dispatch","seq":159,"time":1785052799923,"data":{"parentCallId":"call_00_R6g9Uzx4h0jeUv9g3fno7490","subCallId":"call_00_R6g9Uzx4h0jeUv9g3fno7490:code:1","name":"bash","arguments":{"command":"seq 1 200 | awk '{printf \"line %04d: the quick brown fox jumps over the lazy dog\\n\", $1}'","description":"Generate 200 lines of text"},"isError":false,"content":[{"type":"text","text":"line 0001: the quick brown fox jumps over the lazy dog\nline 0002: the quick brown fox jumps over the lazy dog\nline 0003: the quick brown fox jumps over the lazy dog\nline 0004: the quick s over the lazy dog\nline 0198: the quick brown fox jumps over the lazy dog\nline 0199: the quick brown fox jumps over the lazy dog\nline 0200: the quick brown fox jumps over the lazy dog\n\n\n(Omitted 10629 bytes. Full formatted result stored at: {{cwd}}/.spill/session-2d2b9e84a250/825a63550249-bash.txt. Use read with offset/limit, or grep this path to search within it.)"}]}} {"type":"tool/result","seq":160,"time":1785052799925,"data":{"turn":1,"step":1,"callId":"call_00_R6g9Uzx4h0jeUv9g3fno7490","content":[{"type":"text","text":"200"}],"isError":false},"sourceEventSeqs":[157],"surfaceOp":"append"} {"type":"step/end","seq":161,"time":1785052799926,"data":{"turn":1,"step":1}} {"type":"step/start","seq":162,"time":1785052799928,"data":{"turn":1,"step":2}} diff --git a/examples/tui-agent/tests/snapshots/code-mode/session.jsonl b/examples/tui-agent/tests/snapshots/code-mode/session.jsonl index 858af9a595..7af580f60a 100644 --- a/examples/tui-agent/tests/snapshots/code-mode/session.jsonl +++ b/examples/tui-agent/tests/snapshots/code-mode/session.jsonl @@ -1,4 +1,4 @@ -{"type":"session","version":0,"id":"main-session","createdAt":1785014512062,"cwd":"/tmp/dsh-tui-snapshot-code-mode-7FNsKO"} +{"type":"session","version":0,"id":"main-session","createdAt":1785014512062,"cwd":"{{cwd}}"} {"type":"turn/start","seq":0,"time":1785014512139,"data":{"turn":1,"trigger":{"kind":"message","source":{"kind":"user"}}}} {"type":"user/message","seq":1,"time":1785014512140,"data":{"content":[{"type":"text","text":"Using ONE run_code program: call the bash tool twice — exactly `echo CODE_ONE` then exactly `echo CODE_TWO`. Inside that same program, console.log exactly `captured output`, then return the two outputs joined with a plus sign. Reply with that joined string only and stop."}],"source":{"kind":"user"}},"surfaceOp":"append"} {"type":"session/title","seq":2,"time":1785014512146,"data":{"title":"Using ONE run_code program: call","messageSeqs":[1],"source":{"kind":"fallback"}}} diff --git a/examples/tui-agent/tests/snapshots/cordis-dynamic-toolchain/session.1.jsonl b/examples/tui-agent/tests/snapshots/cordis-dynamic-toolchain/session.1.jsonl index 26519d458e..7027bd50f3 100644 --- a/examples/tui-agent/tests/snapshots/cordis-dynamic-toolchain/session.1.jsonl +++ b/examples/tui-agent/tests/snapshots/cordis-dynamic-toolchain/session.1.jsonl @@ -1,4 +1,4 @@ -{"type":"session","version":0,"id":"22222222-2222-4222-8222-222222222222","createdAt":1783950001000,"cwd":"/tmp/advanced-acp","parentSession":"11111111-1111-4111-8111-111111111111","delegationDepth":1} +{"type":"session","version":0,"id":"22222222-2222-4222-8222-222222222222","createdAt":1783950001000,"cwd":"{{cwd}}","parentSession":"11111111-1111-4111-8111-111111111111","delegationDepth":1} {"type":"turn/start","seq":0,"time":1783957884563,"data":{"turn":1,"trigger":{"kind":"message","source":{"kind":"user"}}}} {"type":"user/message","seq":1,"time":1783957884563,"data":{"content":[{"type":"text","text":"Reply with exactly DIRECT_CHILD_OK and nothing else."}],"source":{"kind":"user"}},"surfaceOp":"append"} {"type":"step/start","seq":2,"time":1783957884564,"data":{"turn":1,"step":1}} diff --git a/examples/tui-agent/tests/snapshots/cordis-dynamic-toolchain/session.2.jsonl b/examples/tui-agent/tests/snapshots/cordis-dynamic-toolchain/session.2.jsonl index 9daa8958dc..5d383f6421 100644 --- a/examples/tui-agent/tests/snapshots/cordis-dynamic-toolchain/session.2.jsonl +++ b/examples/tui-agent/tests/snapshots/cordis-dynamic-toolchain/session.2.jsonl @@ -1,4 +1,4 @@ -{"type":"session","version":0,"id":"33333333-3333-4333-8333-333333333333","createdAt":1783950002000,"cwd":"/tmp/advanced-acp","parentSession":"11111111-1111-4111-8111-111111111111","delegationDepth":1} +{"type":"session","version":0,"id":"33333333-3333-4333-8333-333333333333","createdAt":1783950002000,"cwd":"{{cwd}}","parentSession":"11111111-1111-4111-8111-111111111111","delegationDepth":1} {"type":"turn/start","seq":0,"time":1783957884700,"data":{"turn":1,"trigger":{"kind":"message","source":{"kind":"user"}}}} {"type":"user/message","seq":1,"time":1783957884700,"data":{"content":[{"type":"text","text":"Reply with exactly WORKFLOW_CHILD_OK and nothing else."}],"source":{"kind":"user"}},"surfaceOp":"append"} {"type":"step/start","seq":2,"time":1783957884700,"data":{"turn":1,"step":1}} diff --git a/examples/tui-agent/tests/snapshots/cordis-dynamic-toolchain/session.jsonl b/examples/tui-agent/tests/snapshots/cordis-dynamic-toolchain/session.jsonl index b9ce1d12da..0e9355b4c5 100644 --- a/examples/tui-agent/tests/snapshots/cordis-dynamic-toolchain/session.jsonl +++ b/examples/tui-agent/tests/snapshots/cordis-dynamic-toolchain/session.jsonl @@ -1,4 +1,4 @@ -{"type": "session", "version": 0, "id": "11111111-1111-4111-8111-111111111111", "createdAt": 1783950000000, "cwd": "/tmp/advanced-acp", "delegationDepth": 0} +{"type":"session","version":0,"id":"11111111-1111-4111-8111-111111111111","createdAt":1783950000000,"cwd":"{{cwd}}","delegationDepth":0} {"type":"turn/start","seq":0,"time":1783957884479,"data":{"turn":1,"trigger":{"kind":"message","source":{"kind":"user"}}}} {"type":"user/message","seq":1,"time":1783957884479,"data":{"content":[{"type":"text","text":"Run this advanced flow exactly once: try a no-op temporary Cordis Plugin named snapshot-marker; use run_code to inspect the live temporary Plugins through tools.cordis_inspect; delegate once to a direct spawn child; run one workflow that delegates to another spawn child; stop dyn-1; then reply with exactly ADVANCED_ACP_OK."}],"source":{"kind":"user"}},"surfaceOp":"append"} {"type":"step/start","seq":2,"time":1783957884486,"data":{"turn":1,"step":1}} diff --git a/examples/tui-agent/tests/snapshots/dynamic-workflow/session.1.jsonl b/examples/tui-agent/tests/snapshots/dynamic-workflow/session.1.jsonl index e1dd4a461a..42b73e2084 100644 --- a/examples/tui-agent/tests/snapshots/dynamic-workflow/session.1.jsonl +++ b/examples/tui-agent/tests/snapshots/dynamic-workflow/session.1.jsonl @@ -1,4 +1,4 @@ -{"type":"session","version":0,"id":"583a4db2-3350-436c-b4a5-5615fd159052","createdAt":1783600636316,"cwd":"/var/folders/bn/vj1dvck95yd5jh3x4wskflxm0000gn/T/acp-snap-cwd-vdJYjz","parentSession":"3fd7d599-56b1-493a-930d-f1fc5e1556e8","delegationDepth":1} +{"type":"session","version":0,"id":"583a4db2-3350-436c-b4a5-5615fd159052","createdAt":1783600636316,"cwd":"{{cwd}}","parentSession":"3fd7d599-56b1-493a-930d-f1fc5e1556e8","delegationDepth":1} {"type":"turn/start","seq":0,"time":1783600636316,"data":{"turn":1,"trigger":{"kind":"message","source":{"kind":"user"}}}} {"type":"user/message","seq":1,"time":1783600636316,"data":{"content":[{"type":"text","text":"Reply with exactly the word WF_CHILD_OK and nothing else."}],"source":{"kind":"user"}},"surfaceOp":"append"} {"type":"step/start","seq":2,"time":1783600636316,"data":{"turn":1,"step":1}} diff --git a/examples/tui-agent/tests/snapshots/dynamic-workflow/session.jsonl b/examples/tui-agent/tests/snapshots/dynamic-workflow/session.jsonl index 71bad8720d..9083060639 100644 --- a/examples/tui-agent/tests/snapshots/dynamic-workflow/session.jsonl +++ b/examples/tui-agent/tests/snapshots/dynamic-workflow/session.jsonl @@ -1,4 +1,4 @@ -{"type":"session","version":0,"id":"3fd7d599-56b1-493a-930d-f1fc5e1556e8","createdAt":1783600631835,"cwd":"/var/folders/bn/vj1dvck95yd5jh3x4wskflxm0000gn/T/acp-snap-cwd-vdJYjz","delegationDepth":0} +{"type":"session","version":0,"id":"3fd7d599-56b1-493a-930d-f1fc5e1556e8","createdAt":1783600631835,"cwd":"{{cwd}}","delegationDepth":0} {"type":"turn/start","seq":0,"time":1783600631838,"data":{"turn":1,"trigger":{"kind":"message","source":{"kind":"user"}}}} {"type":"user/message","seq":1,"time":1783600631838,"data":{"content":[{"type":"text","text":"Use the workflow tool exactly once, with args omitted, meta set to { \"name\": \"snapshot-flow\", \"description\": \"one child for the snapshot\" }, and this EXACT script body (copy it verbatim):\nphase('Run')\nconst reply = await agent('Reply with exactly the word WF_CHILD_OK and nothing else.')\nreturn { reply }\nAfter the workflow returns, reply with the single word WORKFLOW_DONE and stop. Do not use any other tool."}],"source":{"kind":"user"}},"surfaceOp":"append"} {"type":"step/start","seq":2,"time":1783600631839,"data":{"turn":1,"step":1}} diff --git a/examples/tui-agent/tests/snapshots/multi-turn-conversation/session.jsonl b/examples/tui-agent/tests/snapshots/multi-turn-conversation/session.jsonl index 76b7ceba58..549dc342a7 100644 --- a/examples/tui-agent/tests/snapshots/multi-turn-conversation/session.jsonl +++ b/examples/tui-agent/tests/snapshots/multi-turn-conversation/session.jsonl @@ -1,4 +1,4 @@ -{"type":"session","version":0,"id":"228b7b82-84ed-49b7-a567-981c03b28c77","createdAt":1783352113760,"cwd":"/tmp/acp-snap-cwd-aN2GRR","delegationDepth":0} +{"type":"session","version":0,"id":"228b7b82-84ed-49b7-a567-981c03b28c77","createdAt":1783352113760,"cwd":"{{cwd}}","delegationDepth":0} {"type":"turn/start","seq":0,"time":1783352113765,"data":{"turn":1,"trigger":{"kind":"message","source":{"kind":"user"}}}} {"type":"user/message","seq":1,"time":1783352113765,"data":{"content":[{"type":"text","text":"Reply with exactly the word: ONE. No tools."}],"source":{"kind":"user"}},"surfaceOp":"append"} {"type":"step/start","seq":2,"time":1783352113767,"data":{"turn":1,"step":1}} diff --git a/examples/tui-agent/tests/snapshots/multi-turn-conversation/terminal.expected.txt b/examples/tui-agent/tests/snapshots/multi-turn-conversation/terminal.expected.txt index e4fa63468a..09f3cb1fb4 100644 --- a/examples/tui-agent/tests/snapshots/multi-turn-conversation/terminal.expected.txt +++ b/examples/tui-agent/tests/snapshots/multi-turn-conversation/terminal.expected.txt @@ -15,8 +15,8 @@ buffer style 0-2 fg=bright-blue bold underline 5| "Reply with exactly the word: ONE. No tools. " 6| -7| "Entering plan mode (applies from the next step). Use /plan off to leave. " - style 0-71 fg=bright-black +7| "Plan mode on. Use /plan off to leave. " + style 0-36 fg=bright-black 8| 9| "Assistant " style 0-8 fg=bright-magenta bold underline @@ -28,17 +28,17 @@ buffer 13| "Model wait 0.0s · Completed 2026-07-21 12:00:00 " style 0-46 dim 14| -15| "Leaving plan mode (applies from the next step). " - style 0-46 fg=bright-black -16| -17| "You " - style 0-2 fg=bright-blue bold underline -18| "Reply with exactly the word: TWO. No tools. " -19| -20| "Context · plan-mode " +15| "Context · plan-mode " style 0-18 dim -21| "The user switched this session back to the default mode. " +16| "The user switched this session back to the default mode. " style 0-55 fg=bright-black +17| +18| "Plan mode off. " + style 0-13 fg=bright-black +19| +20| "You " + style 0-2 fg=bright-blue bold underline +21| "Reply with exactly the word: TWO. No tools. " 22| 23| "Assistant " style 0-8 fg=bright-magenta bold underline diff --git a/examples/tui-agent/tests/snapshots/todo-plan/session.jsonl b/examples/tui-agent/tests/snapshots/todo-plan/session.jsonl index 878948fd26..3591582b9b 100644 --- a/examples/tui-agent/tests/snapshots/todo-plan/session.jsonl +++ b/examples/tui-agent/tests/snapshots/todo-plan/session.jsonl @@ -1,4 +1,4 @@ -{"type":"session","version":0,"id":"b0f1f758-dcf0-474e-851d-e62c11ec0a09","createdAt":1783352057652,"cwd":"/tmp/acp-snap-cwd-AYilT7","delegationDepth":0} +{"type":"session","version":0,"id":"b0f1f758-dcf0-474e-851d-e62c11ec0a09","createdAt":1783352057652,"cwd":"{{cwd}}","delegationDepth":0} {"type":"turn/start","seq":0,"time":1783352057655,"data":{"turn":1,"trigger":{"kind":"message","source":{"kind":"user"}}}} {"type":"user/message","seq":1,"time":1783352057655,"data":{"content":[{"type":"text","text":"Use the todo_write tool to record a plan with exactly three todos: \"read the code\" (in_progress), \"write the fix\" (pending), \"run the tests\" (pending). Send all three in one todo_write call. Then reply with the single word DONE and stop."}],"source":{"kind":"user"}},"surfaceOp":"append"} {"type":"step/start","seq":2,"time":1783352057657,"data":{"turn":1,"step":1}} diff --git a/examples/tui-agent/tests/tui-keyless-smoke.e2e.ts b/examples/tui-agent/tests/tui-keyless-smoke.e2e.ts index 3908670e27..7ae98167de 100644 --- a/examples/tui-agent/tests/tui-keyless-smoke.e2e.ts +++ b/examples/tui-agent/tests/tui-keyless-smoke.e2e.ts @@ -149,14 +149,14 @@ describe('tui-agent keyless smoke (real Loader tree in a PTY)', () => { actions: [ { waitFor: 'main-session-', send: '/plan' }, { waitFor: '[off|message] — Enter or leave plan mode', send: '\r' }, - { waitFor: 'Entering plan mode (applies from the next step). Use /plan off to leave.', send: '/exit\r' }, + { waitFor: 'Plan mode on. Use /plan off to leave.', send: '/exit\r' }, ], }) expect(output).toContain('DEEPSEEK') expect(output).toContain('HARNESS') expect(output).toContain('main-session-') expect(output).toContain('[off|message] — Enter or leave plan mode') - expect(output).toContain('Entering plan mode (applies from the next step). Use /plan off to leave.') + expect(output).toContain('Plan mode on. Use /plan off to leave.') // Borderless: no box-drawing frame around the banner. expect(output).not.toContain('╭') expect(output).not.toContain('╮') @@ -183,15 +183,15 @@ describe('tui-agent keyless smoke (real Loader tree in a PTY)', () => { // Gating /status on it keeps the assertion race-free; the diagnostics // card is then exercised through the same real Loader/PTY composition. { waitFor: 'scripted session title — DeepSeek Harness', send: '/plan off\r' }, - { waitFor: 'Leaving plan mode (applies from the next step).', send: 'Confirm the scripted run left plan mode.\r' }, + { waitFor: 'Plan mode off.', send: 'Confirm the scripted run left plan mode.\r' }, { waitFor: 'Default mode confirmed.', send: '/status\r' }, { waitFor: 'Session status', send: '/exit\r' }, ], }) expect(output).toContain('I need one decision before I continue.') expect(output).toContain('Reasoning effort: Max.') - expect(output).toContain('Entering plan mode (applies from the next step). Use /plan off to leave.') - expect(output).toContain('Leaving plan mode (applies from the next step).') + expect(output).toContain('Plan mode on. Use /plan off to leave.') + expect(output).toContain('Plan mode off.') expect(output).toContain('Default mode confirmed.') expect(output).toContain(String.raw`\x1b]2;MODEL_CONTROLLED\x07`) expect(output).toContain(String.raw`\x1b[999CMODEL_CURSOR`) diff --git a/examples/tui-agent/tests/tui.snapshot.ts b/examples/tui-agent/tests/tui.snapshot.ts index 98abee3f0a..af7c027d6d 100644 --- a/examples/tui-agent/tests/tui.snapshot.ts +++ b/examples/tui-agent/tests/tui.snapshot.ts @@ -4,7 +4,7 @@ import { basename, dirname, isAbsolute, join, relative, sep } from 'node:path' import { fileURLToPath } from 'node:url' import { afterAll, describe, expect, it, vi } from 'vitest' import { Context } from 'cordis' -import { scrubRequestHeaders } from '@deepseek-ai/dsh-acp-snapshot' +import { scrubRequestHeaders, tokenizeSessionFixtureCwd } from '@deepseek-ai/dsh-acp-snapshot' import type { Agent } from '@deepseek-ai/dsh-agent' import * as AgentCore from '@deepseek-ai/dsh-agent-spine-demo' import { LocalBashExecutor } from '@deepseek-ai/dsh-bash-local' @@ -415,10 +415,16 @@ async function runScenario(scenario: Scenario): Promise { async function writeRecording(scenario: Scenario, result: ScenarioResult): Promise { const dir = scenarioDir(scenario) await mkdir(dir, { recursive: true }) - await writeFile(join(dir, 'session.jsonl'), scrubRequestHeaders(rawSessionLog(result.parent))) + await writeFile( + join(dir, 'session.jsonl'), + scrubRequestHeaders(tokenizeSessionFixtureCwd(rawSessionLog(result.parent))), + ) expect(result.children).toHaveLength(scenario.childSessions ?? 0) for (const [index, child] of result.children.entries()) { - await writeFile(join(dir, `session.${index + 1}.jsonl`), scrubRequestHeaders(rawSessionLog(child))) + await writeFile( + join(dir, `session.${index + 1}.jsonl`), + scrubRequestHeaders(tokenizeSessionFixtureCwd(rawSessionLog(child))), + ) } } diff --git a/knip.json b/knip.json index d010bcf21d..38eb32c223 100644 --- a/knip.json +++ b/knip.json @@ -31,6 +31,7 @@ "entry": [ "headless-agent/tests/fixtures/cli-mock-llm.ts", "headless-agent/tests/fixtures/semantic-checkpoint-agent.ts", + "headless-agent/tests/fixtures/subagent-inheritance-agent.ts", "headless-agent/tests/fixtures/goal-domain/seed-goal.ts", "headless-agent/tests/fixtures/time-context-driver.ts", "headless-agent/tests/fixtures/time-context-mock-llm.ts", @@ -100,6 +101,16 @@ "tests/**/*.tsx" ] }, + "packages/client/ui-goal": { + "entry": [ + "tests/**/*.spec.tsx" + ], + "project": [ + "src/**/*.ts", + "src/**/*.tsx", + "tests/**/*.tsx" + ] + }, "packages/client/web-react": { "entry": [ "tests/**/*.spec.tsx" diff --git a/package.json b/package.json index a0fc4f780f..afa74fdf1e 100644 --- a/package.json +++ b/package.json @@ -14,13 +14,14 @@ "website" ], "scripts": { - "build": "tsc -b && tsdown", + "build": "npm run build:lib && npm run build:web", + "build:lib": "tsc -b && tsdown", "build:web": "pnpm --filter @deepseek-ai/dsh-frontend run build", "clean": "tsx scripts/clean.ts", "change-scope": "tsx scripts/change-scope.ts", "typecheck": "tsc -b", - "lint": "eslint .", - "lint:fix": "eslint . --fix", + "lint": "node --max-old-space-size=8192 node_modules/eslint/bin/eslint.js .", + "lint:fix": "node --max-old-space-size=8192 node_modules/eslint/bin/eslint.js . --fix", "duplication": "jscpd --config .jscpd.json packages scripts", "test": "vitest run", "test:coverage": "vitest run --coverage", @@ -29,7 +30,7 @@ "test:snapshot:record": "DSH_SNAPSHOT=record vitest run --config vitest.snapshot.config.ts --update", "test:snapshot:refresh": "DSH_SNAPSHOT=refresh vitest run --config vitest.snapshot.config.ts", "migrate:packed-session-fixtures": "tsx scripts/migrate-packed-session-fixtures.ts", - "test:web": "npm run build && npm run build:web && vitest run --config vitest.web.config.ts", + "test:web": "npm run build && vitest run --config vitest.web.config.ts", "test:gui": "vitest run packages/client packages/host", "check:all": "tsx scripts/run-gates.ts check-all", "check:ci": "tsx scripts/run-gates.ts ci-primary", @@ -95,13 +96,13 @@ "constraints": "tsx scripts/check-workspace-constraints.ts", "doc-sync": "tsx scripts/run-gates.ts doc-sync", "hygiene": "pnpm run knip && pnpm run publint && pnpm run constraints && pnpm run verify-package-invariants && pnpm run verify-built-package-invariants && pnpm run verify-cordis-config && pnpm run verify-node-next-types && pnpm run verify-runtime-closure", - "dsh": "node --import tsx apps/cli/src/bin.ts", + "dsh": "node --experimental-transform-types --import ./scripts/tspath-loader.ts apps/cli/src/bin.ts", "demo:headless": "node --import tsx packages/examples/cli-demo/src/bin.ts --config examples/headless-agent/cordis.yml", - "demo:tui": "node --import tsx apps/cli/src/bin.ts", + "demo:tui": "node --experimental-transform-types --import ./scripts/tspath-loader.ts apps/cli/src/bin.ts", "demo:code-mode": "node scripts/demo-code-mode.mjs", "demo:cordis": "node scripts/demo-cordis.mjs", "demo:acp": "node --import tsx packages/examples/acp-demo/src/bin.ts --config examples/acp-agent/cordis.yml", - "demo:web": "npm run build && npm run build:web && node --import tsx apps/cli/src/bin.ts web", + "demo:web": "npm run build && node --experimental-transform-types --import ./scripts/tspath-loader.ts apps/cli/src/bin.ts web", "mock:llm": "node --import tsx packages/support/llm-mock-server/src/bin.ts", "dev:web": "tsx scripts/dev-web.ts --poll", "postinstall": "node scripts/install-lefthook.mjs" diff --git a/packages/client/connection/src/client/api.ts b/packages/client/connection/src/client/api.ts index 77f5445f6d..978d4c3378 100644 --- a/packages/client/connection/src/client/api.ts +++ b/packages/client/connection/src/client/api.ts @@ -12,6 +12,7 @@ export type { CommandsApi, CommandDescriptor, SkillsApi, SkillEntry, ModelCatalogFailure, ModelCatalogModel, ModelProviderGroup, ModelReasoning, ModelReasoningEffort, ModelTarget, SessionModels, + GoalsApi, GoalRef, } from '@deepseek-ai/dsh-host-apiproxy/api' export type { ToolCallView, ToolResultView } from '@deepseek-ai/dsh-tools/presentation' export type { diff --git a/packages/client/connection/src/client/fixture.ts b/packages/client/connection/src/client/fixture.ts index 4bb5d2baa3..bcb88ac0f7 100644 --- a/packages/client/connection/src/client/fixture.ts +++ b/packages/client/connection/src/client/fixture.ts @@ -302,6 +302,34 @@ function viewFor(event: SessionEvent, log: readonly SessionEvent[]): ToolEventVi return undefined } +/** + * Fixture parallel of the plan unit's double-event fold: `command/run` + * records named `plan` set the wanted target (`off` → false, else true); + * `plan/mode` commits and clears it. `wanted` is exposed for the prompt + * boundary (the fixture's agent/step parallel). + */ +function foldPlan(log: readonly SessionEvent[]): { active: boolean; pending: boolean; wanted: boolean | null } { + let active = false + let wanted: boolean | null = null + for (const event of log) { + const item = event as unknown as { type: string; data?: Record } + if (item.type === 'command/run' && item.data?.['name'] === 'plan') { + const args = item.data['args'] + wanted = (typeof args === 'string' ? args : '').trim() !== 'off' + } else if (item.type === 'plan/mode') { + active = item.data?.['active'] === true + wanted = null + } + } + return { active, pending: wanted !== null && wanted !== active, wanted } +} + +/** The plan projection's wire view over the full log. */ +function planViewOf(log: readonly SessionEvent[]): { active: boolean; pending: boolean } { + const plan = foldPlan(log) + return { active: plan.active, pending: plan.pending } +} + /** Fixture parallel of the host's projection units: whole current values per key over the full log. */ function projectionValuesOf(log: readonly SessionEvent[]): Record { const values: Record = {} @@ -311,6 +339,10 @@ function projectionValuesOf(log: readonly SessionEvent[]): Record= 0; i--) { + const event = log[i] as unknown as { + type: string + data?: { source?: { kind?: string; round?: number; change?: FxGoalChange } } + } | undefined + if (event === undefined || event.type !== 'user/message') continue + const source = event.data?.source + if (source?.kind !== 'goal' || source.round !== 0) continue + const change = source.change + // eslint-disable-next-line @typescript-eslint/no-unnecessary-condition + if (change === undefined || change.kind !== 'goal/change') continue + if (change.operation === 'clear') return null + return { goal: change.goal, roundsStarted: change.roundsStarted, createdAt: change.createdAt, updatedAt: change.updatedAt } + } + return null +} + interface StreamConn { push(envelope: RpcRequest): void } @@ -572,6 +672,47 @@ export function createFixtureApi(options: FixtureOptions = {}): ApiProxy { for (const frame of projectionFramesOf(id, log, event)) emitMux(frame) } + /** Append one goal/change as its round-zero goal-sourced user message (host GoalService parallel). */ + const appendGoalChange = (id: SessionId, change: FxGoalChange): FxGoalProjection => { + const ref = change.operation === 'clear' ? change.cleared : change.goal + const payload = change.operation === 'clear' + ? { cleared: change.cleared, clearedAt: change.clearedAt } + : { goal: change.goal, roundsStarted: change.roundsStarted, createdAt: change.createdAt, updatedAt: change.updatedAt } + append(id, { + type: 'user/message', surfaceOp: 'append', + data: userMessage( + text(`${JSON.stringify(payload)}`), + { kind: 'goal', goalId: ref.id, revision: ref.revision, round: 0, change } as unknown as MessageSource, + ), + }) + return backscanGoal(logOf(id)) as FxGoalProjection + } + + /** Shared CAS mutation path of the goal verbs (undefined next = invalid transition). */ + const fxMutateGoal = ( + request: RpcRequest<{ sessionId: SessionId; ref: { id: string; revision: number } }>, + ref: { id: string; revision: number }, + next: (current: FxGoalProjection) => FxGoalProjection['goal'] | undefined, + ): Promise> => { + const missing = requireSession(request) + if (missing !== undefined) return missing + const id = request.payload.sessionId + const current = backscanGoal(logOf(id)) + if (current === null || current.goal.id !== ref.id || current.goal.revision !== ref.revision) { + return err(request, { code: 'internal', message: 'stale or missing goal revision', details: { goalCode: 'GOAL_STALE_REVISION' } }) + } + const goal = next(current) + if (goal === undefined) { + return err(request, { code: 'internal', message: `invalid goal transition from "${current.goal.phase}"`, details: { goalCode: 'GOAL_INVALID_TRANSITION' } }) + } + const projection = appendGoalChange(id, { + kind: 'goal/change', version: 1, + operation: goal.phase === current.goal.phase ? 'edit' : goal.phase === 'paused' ? 'pause' : goal.phase === 'active' ? 'resume' : 'complete', + goal, roundsStarted: current.roundsStarted, createdAt: current.createdAt, updatedAt: Date.now(), + }) + return ok(request, { ref: { id: projection.goal.id as never, revision: projection.goal.revision } }) + } + /** At most one in-flight replay per session; cancel clears it. */ const replays = new Map; finish(aborted: boolean): void }>() @@ -804,6 +945,12 @@ export function createFixtureApi(options: FixtureOptions = {}): ApiProxy { nextTurn.set(id, turn + 1) setRunning(id, true) append(id, { type: 'turn/start', data: { turn, trigger: { kind: 'message', source: { kind: 'user' } } } }) + // Boundary flush parallel (the host's agent/step seam): an outstanding + // /plan selection commits as plan/mode inside the opened turn. + const plan = foldPlan(logOf(id)) + if (plan.wanted !== null && plan.wanted !== plan.active) { + append(id, { type: 'plan/mode', data: { active: plan.wanted } }) + } append(id, { type: 'user/message', surfaceOp: 'append', data: userMessage(content) }) startReply( id, @@ -934,7 +1081,8 @@ export function createFixtureApi(options: FixtureOptions = {}): ApiProxy { commands: [ { name: 'compact', description: 'fixture：压缩当前会话上下文' }, { name: 'echo', description: 'fixture：回显参数', input: { hint: 'text to echo' } }, - { name: 'goal-fixture', description: 'fixture：目标样本命令', input: { hint: '' } }, + { name: 'goal', description: 'set or view the goal for a long-running task', input: { hint: '' } }, + { name: 'plan', description: 'Enter or leave plan mode', input: { hint: '[off|message]' } }, ], }) }, @@ -950,15 +1098,52 @@ export function createFixtureApi(options: FixtureOptions = {}): ApiProxy { const match = /^\/(\S+)((?:\s.*)?)$/.exec(request.payload.line.trim()) const name = match?.[1] const args = match?.[2] ?? '' + if (name === 'goal') { + // Host parallel: /goal with an objective creates (or reports) the + // current goal; the command lifecycle pair brackets the mutation. + const commandId = `fx-cmd-${logOf(id).length}` as CommandId + append(id, { type: 'command/run', data: { commandId, name, args, source: { kind: 'user' } } }) + const objective = args.trim() + const current = backscanGoal(logOf(id)) + let text: string + if (objective === '') { + text = current === null ? 'No goal is set. Usage: /goal ' : `Current goal: ${current.goal.objective}` + } else if (current !== null && current.goal.phase !== 'complete') { + text = `A goal already exists (${current.goal.objective}). Clear it first.` + } else { + const created = appendGoalChange(id, { + kind: 'goal/change', version: 1, operation: 'create', + goal: { id: `fx-goal-${logOf(id).length}`, revision: 1, objective, phase: 'active', maxGoalRounds: 256 }, + roundsStarted: 0, createdAt: Date.now(), updatedAt: Date.now(), + }) + text = `Goal created: ${created.goal.objective}` + } + append(id, { type: 'command/done', data: { commandId, kind: 'success', text } }) + return ok(request, { matched: true as const, commandId }) + } + // Host parallel: /plan on an idle fixture session commits plan/mode + // immediately (the boundary flush covers only a running turn), so the + // outcome copy matches the immediate branch of the host handler. + const running = summaryOf(id)?.running === true const outcomes: Record = { compact: 'fixture：已压缩（假动作）', echo: args.trim(), - 'goal-fixture': `fixture：goal 已设置（${id}）`, + plan: args.trim() === 'off' + ? (running ? 'Leaving plan mode (applies from the next step).' : 'Plan mode off.') + : (running + ? 'Entering plan mode (applies from the next step). Use /plan off to leave.' + : 'Plan mode on. Use /plan off to leave.'), } const text = name === undefined ? undefined : outcomes[name] if (name === undefined || text === undefined) return ok(request, { matched: false as const }) const commandId = `fx-cmd-${logOf(id).length}` as CommandId append(id, { type: 'command/run', data: { commandId, name, args, source: { kind: 'user' } } }) + if (name === 'plan' && !running) { + const plan = foldPlan(logOf(id)) + if (plan.wanted !== null && plan.wanted !== plan.active) { + append(id, { type: 'plan/mode', data: { active: plan.wanted } }) + } + } append(id, { type: 'command/done', data: { commandId, kind: 'success', ...text === '' ? {} : { text } } }) return ok(request, { matched: true as const, commandId }) }, @@ -974,6 +1159,62 @@ export function createFixtureApi(options: FixtureOptions = {}): ApiProxy { }) }, }, + goals: { + // Mutation-only mirror of the host handlers: each verb CAS-checks the + // projected current goal, appends the whole-value change (the mux + // stream and projection frame ride the shared append path), and + // acknowledges with the new ref only. + create: (request) => { + const missing = requireSession(request) + if (missing !== undefined) return missing + const id = request.payload.sessionId + const current = backscanGoal(logOf(id)) + if (current !== null && current.goal.phase !== 'complete') { + return err(request, { code: 'internal', message: `goal "${current.goal.id}" already exists`, details: { goalCode: 'GOAL_ALREADY_EXISTS' } }) + } + const projection = appendGoalChange(id, { + kind: 'goal/change', version: 1, operation: 'create', + goal: { id: `fx-goal-${logOf(id).length}`, revision: 1, objective: request.payload.objective, phase: 'active', maxGoalRounds: request.payload.maxGoalRounds ?? 256 }, + roundsStarted: 0, createdAt: Date.now(), updatedAt: Date.now(), + }) + return ok(request, { ref: { id: projection.goal.id as never, revision: projection.goal.revision } }) + }, + edit: request => fxMutateGoal(request, request.payload.ref, current => ({ + ...current.goal, + revision: current.goal.revision + 1, + ...request.payload.objective === undefined ? {} : { objective: request.payload.objective }, + ...request.payload.maxGoalRounds === undefined ? {} : { maxGoalRounds: request.payload.maxGoalRounds }, + })), + pause: request => fxMutateGoal(request, request.payload.ref, current => ( + current.goal.phase === 'active' + ? { ...current.goal, revision: current.goal.revision + 1, phase: 'paused' } + : undefined + )), + resume: request => fxMutateGoal(request, request.payload.ref, current => ( + current.goal.phase === 'paused' || current.goal.phase === 'blocked' || current.goal.phase === 'active' + ? { ...current.goal, revision: current.goal.revision + 1, phase: 'active' } + : undefined + )), + complete: request => fxMutateGoal(request, request.payload.ref, current => ( + current.goal.phase === 'complete' + ? undefined + : { ...current.goal, revision: current.goal.revision + 1, phase: 'complete' } + )), + clear: (request) => { + const missing = requireSession(request) + if (missing !== undefined) return missing + const id = request.payload.sessionId + const current = backscanGoal(logOf(id)) + if (current === null || current.goal.id !== request.payload.ref.id || current.goal.revision !== request.payload.ref.revision) { + return err(request, { code: 'internal', message: 'stale or missing goal revision', details: { goalCode: 'GOAL_STALE_REVISION' } }) + } + appendGoalChange(id, { + kind: 'goal/change', version: 1, operation: 'clear', + cleared: { id: current.goal.id, revision: current.goal.revision + 1 }, clearedAt: Date.now(), + }) + return ok(request, { cleared: true as const }) + }, + }, events: { async *mux(_request, signal) { const conn = new FxInbox() @@ -1104,6 +1345,12 @@ export class FixtureApiClient extends AbstractApiClient { // The in-memory execute never blocks, so a never-aborting signal is faithful here. case 'command.execute': return this.api.commands.execute(request, new AbortController().signal) case 'skill.list': return this.api.skills.list(request) + case 'goal.create': return this.api.goals.create(request) + case 'goal.edit': return this.api.goals.edit(request) + case 'goal.pause': return this.api.goals.pause(request) + case 'goal.resume': return this.api.goals.resume(request) + case 'goal.complete': return this.api.goals.complete(request) + case 'goal.clear': return this.api.goals.clear(request) } } diff --git a/packages/client/connection/src/client/index.ts b/packages/client/connection/src/client/index.ts index 1ce30dbdc4..53e5b2bc3f 100644 --- a/packages/client/connection/src/client/index.ts +++ b/packages/client/connection/src/client/index.ts @@ -20,6 +20,7 @@ export type { RpcRequest, RpcResponse, RpcResult, RpcError, RpcErrorCode, ClientRequest, ServerResponse, ServerRequest, ClientResponse, RpcMessage, RpcReceipt, IApiClient, SessionId, SessionEvent, ContentBlock, StreamChunk, + GoalsApi, GoalRef, } from './api.ts' export { RpcId, AbstractApiClient, transportError } from './api.ts' diff --git a/packages/client/connection/tests/fake-api.ts b/packages/client/connection/tests/fake-api.ts index 1dac997ae2..d974c4bf3b 100644 --- a/packages/client/connection/tests/fake-api.ts +++ b/packages/client/connection/tests/fake-api.ts @@ -127,6 +127,15 @@ export class FakeApiClient implements IApiClient { list: (payload: unknown) => this.record('skill.list', payload, this.onSkillList(payload)), } + readonly goals: IApiClient['goals'] = { + create: payload => this.record('goal.create', payload, Promise.resolve(ok({ ref: { id: 'fake-goal' as never, revision: 1 } }))), + edit: payload => this.record('goal.edit', payload, Promise.resolve(ok({ ref: { id: 'fake-goal' as never, revision: 1 } }))), + pause: payload => this.record('goal.pause', payload, Promise.resolve(ok({ ref: { id: 'fake-goal' as never, revision: 1 } }))), + resume: payload => this.record('goal.resume', payload, Promise.resolve(ok({ ref: { id: 'fake-goal' as never, revision: 1 } }))), + complete: payload => this.record('goal.complete', payload, Promise.resolve(ok({ ref: { id: 'fake-goal' as never, revision: 1 } }))), + clear: payload => this.record('goal.clear', payload, Promise.resolve(ok({ cleared: true as const }))), + } + /** When true, streams never fire onOpen (misbehaving-carrier material for the handshake timeout guard). */ suppressStreamOpen = false diff --git a/packages/client/connection/tests/fixture-commands.spec.ts b/packages/client/connection/tests/fixture-commands.spec.ts index bd66d124a4..f589024a24 100644 --- a/packages/client/connection/tests/fixture-commands.spec.ts +++ b/packages/client/connection/tests/fixture-commands.spec.ts @@ -23,7 +23,7 @@ describe('createFixtureApi commands/skills', () => { expect(response.rpcId).toBe(request.rpcId) if (!response.result.ok) throw new Error('list failed') const commands = response.result.value.commands - expect(commands.map(c => c.name)).toEqual(['compact', 'echo', 'goal-fixture']) + expect(commands.map(c => c.name)).toEqual(['compact', 'echo', 'goal', 'plan']) // input hint rides only the commands declaring it. const echo = commands.find(c => c.name === 'echo') expect(echo?.input?.hint).toBeTruthy() @@ -64,11 +64,11 @@ describe('createFixtureApi commands/skills', () => { it('addresses execute to the session; an unknown session errs', async () => { const api = createFixtureApi() - const hit = await api.commands.execute(req({ sessionId: sid('fx-alpha'), line: '/goal-fixture ship' }), signal) + const hit = await api.commands.execute(req({ sessionId: sid('fx-alpha'), line: '/goal ship' }), signal) if (!hit.result.ok) throw new Error('execute failed') expect(hit.result.value.matched).toBe(true) - const missing = await api.commands.execute(req({ sessionId: sid('fx-nope'), line: '/goal-fixture ship' }), signal) + const missing = await api.commands.execute(req({ sessionId: sid('fx-nope'), line: '/goal ship' }), signal) expect(missing.result).toMatchObject({ ok: false, error: { code: 'session-not-found' } }) }) diff --git a/packages/client/connection/tests/fixture.spec.ts b/packages/client/connection/tests/fixture.spec.ts index 2d8015506f..8f51861283 100644 --- a/packages/client/connection/tests/fixture.spec.ts +++ b/packages/client/connection/tests/fixture.spec.ts @@ -69,9 +69,11 @@ describe('createFixtureApi', () => { // tail block still rides it — empty-log cut at -1, the host convention. const empty = await api.sessions.history(req({ sessionId: sid('no-such'), maxMessages: 10 })) if (!empty.result.ok) throw new Error('empty failed') - // Fixture composes the todos unit (host parallel when tool-todo is mounted): null before any write. + // Fixture composes the todos + plan units (host parallel when tool-todo + // and plan-mode are mounted): the empty-log values. expect(empty.result.value).toEqual({ - events: [], hasMore: false, projections: { asOfSeq: -1, values: { todos: null } }, + events: [], hasMore: false, + projections: { asOfSeq: -1, values: { goal: null, todos: null, plan: { active: false, pending: false } } }, }) }) @@ -206,7 +208,7 @@ describe('createFixtureApi', () => { const envelopes: RpcRequest[] = [] for await (const envelope of api.events.mux(req({}), abort.signal)) { envelopes.push(envelope) - if (envelopes.length >= 4) abort.abort() + if (envelopes.length >= 7) abort.abort() } return envelopes } @@ -214,13 +216,15 @@ describe('createFixtureApi', () => { const second = await openOnce() expect(first[0]?.payload).toMatchObject({ type: 'session/subscribed', sessionId: 'fx-alpha' }) expect((first[0]?.payload as { lastSeq: number }).lastSeq).toBeGreaterThan(0) - // Projection baseline frames follow the subscribed frame (title + todos units). + // Projection baseline frames follow the subscribed frame (title + todos + plan + goal units). expect(first[1]?.payload).toMatchObject({ type: 'session/projection', sessionId: 'fx-alpha', key: 'title', value: 'Fixture 历史会话' }) expect(first[2]?.payload).toMatchObject({ type: 'session/projection', sessionId: 'fx-alpha', key: 'todos' }) - expect(first[3]?.payload).toMatchObject({ type: 'approval/requested', toolName: 'dangerous_tool' }) - expect(second[3]?.rpcId).toBe(first[3]?.rpcId) // stable rpcId across replays (host replay semantics) - expect(first[4]?.payload).toMatchObject({ type: 'question/requested', sessionId: 'fx-alpha' }) - expect(second[4]?.rpcId).toBe(first[4]?.rpcId) + expect(first[3]?.payload).toMatchObject({ type: 'session/projection', sessionId: 'fx-alpha', key: 'plan', value: { active: false, pending: false } }) + expect(first[4]?.payload).toMatchObject({ type: 'session/projection', sessionId: 'fx-alpha', key: 'goal', value: null }) + expect(first[5]?.payload).toMatchObject({ type: 'approval/requested', toolName: 'dangerous_tool' }) + expect(second[5]?.rpcId).toBe(first[5]?.rpcId) // stable rpcId across replays (host replay semantics) + expect(first[6]?.payload).toMatchObject({ type: 'question/requested', sessionId: 'fx-alpha' }) + expect(second[6]?.rpcId).toBe(first[6]?.rpcId) }) it('steer with no replay in flight falls through to a fresh queued turn; non-text blocks stringify empty', async () => { @@ -698,6 +702,29 @@ describe('FixtureApiClient (protocol-level fake carrier)', () => { const moved = await client.workspace.insertSessionBefore({ workspaceId: wsid, sessionId: attached.result.value.sessionId }) if (!moved.result.ok) throw new Error('workspace move failed') expect(moved.result.value.workspace.sessionIds).toEqual([attached.result.value.sessionId]) + // Goal lifecycle over the fixture fold: create → edit → pause → resume → complete → clear; + // every mutation acknowledges with the NEW CAS ref (state rides the projection frames). + const goalCreated = await client.goals.create({ sessionId: id, objective: 'ship it' }) + if (!goalCreated.result.ok) throw new Error('goal create failed') + let ref = goalCreated.result.value.ref + expect(ref.revision).toBe(1) + const edited = await client.goals.edit({ sessionId: id, ref, objective: 'ship it v2' }) + if (!edited.result.ok) throw new Error('goal edit failed') + ref = edited.result.value.ref + const paused = await client.goals.pause({ sessionId: id, ref }) + if (!paused.result.ok) throw new Error('goal pause failed') + ref = paused.result.value.ref + const resumed = await client.goals.resume({ sessionId: id, ref }) + if (!resumed.result.ok) throw new Error('goal resume failed') + ref = resumed.result.value.ref + // A stale ref loses the CAS check. + expect((await client.goals.pause({ sessionId: id, ref: { ...ref, revision: 1 } })).result.ok).toBe(false) + const completed = await client.goals.complete({ sessionId: id, ref }) + if (!completed.result.ok) throw new Error('goal complete failed') + ref = completed.result.value.ref + // complete → complete is an invalid transition. + expect((await client.goals.complete({ sessionId: id, ref })).result.ok).toBe(false) + expect((await client.goals.clear({ sessionId: id, ref })).result).toEqual({ ok: true, value: { cleared: true } }) }) it('maps empty, prompt-reject, and workspace-first query scenarios', async () => { diff --git a/packages/client/runtime/src/client/contract/session.ts b/packages/client/runtime/src/client/contract/session.ts new file mode 100644 index 0000000000..577b18ad77 --- /dev/null +++ b/packages/client/runtime/src/client/contract/session.ts @@ -0,0 +1,63 @@ +/** + * The outward session face. Feature packages never see the concrete Session + * class: components read conversation state through `useSession` (the + * ObservableSnapshot half), and orchestration code calls the behavior verbs + * below — nothing else. Widening this interface is the explicit act of + * widening what features may do to a session (and what every test fixture + * must stub); runtime-internal entry points (history staging, wire-frame + * dispatch) stay on the class, invisible out here. + */ +import type { ContentBlock } from '@deepseek-ai/dsh-llm/types' +import type { RpcResult, SessionId } from '@deepseek-ai/dsh-client-connection/client' +import type { ConversationSnapshot } from '../sessions/conversation.ts' +import type { ObservableSnapshot } from './store.ts' + +/** Key-addressed projection read face (the useProjection resolution path; see ProjectionValueStore). */ +export interface ProjectionsFace { + /** + * The identity-stable bare observable for one projection key (absence is + * an `undefined` snapshot, never a missing face). + * @param key - projection key. + * @returns the key's value face. + */ + faceOf(key: string): ObservableSnapshot +} + +/** Identity plus the behavior verbs features may invoke on a session. */ +export interface ISession { + /** The session's host identity (agent id — same axis). */ + readonly sessionId: SessionId + /** Host-computed projection values by key (the useProjection seat). */ + readonly projections: ProjectionsFace + /** + * Send a prompt into the session. + * @param content - model-facing content blocks. + * @param mode - 'queue' appends a turn; 'steer' interrupts the running one. + * @returns acceptance, or the business error (also mirrored into snapshot.promptError). + */ + prompt(content: ContentBlock[], mode: 'queue' | 'steer'): Promise> + /** + * Cancel the running turn. + * @returns acceptance, or the business error. + */ + cancel(): Promise> + /** + * Extend the history window backwards (older messages pagination). + * @returns completion; failures land in snapshot.openState/loadingOlder. + */ + loadOlder(): Promise + /** + * Exhaust the available history for inspection features. + * An abort stops before the next page without abandoning an active request. + * @param signal - Consumer lifetime; abort is observed between pages. + * @returns completion when history is exhausted or paging stops making progress. + */ + loadAllHistory(signal?: AbortSignal): Promise +} + +/** + * The full outward face: behavior verbs plus the conversation read side + * (the `useSession` hook source). This is the type carried by + * `SessionBinding.session` and the provide channel. + */ +export type SessionFace = ISession & ObservableSnapshot diff --git a/packages/client/runtime/src/client/contract/sessions-port.ts b/packages/client/runtime/src/client/contract/sessions-port.ts new file mode 100644 index 0000000000..466e26fe12 --- /dev/null +++ b/packages/client/runtime/src/client/contract/sessions-port.ts @@ -0,0 +1,47 @@ +/** + * Cross-domain sessions face: the contract surface sibling domains (today: + * workspaces) consume instead of the sessions implementation. The sessions + * domain satisfies it structurally — SessionsService is assignable, checked + * wherever the assembly layer or a test injects the real service — so + * widening this face is the explicit act of widening the inter-domain + * dependency. + */ + +import type { SessionId, WorkspaceId } from '@deepseek-ai/dsh-client-connection/client' +import type { ObservableSnapshot } from './store.ts' + +/** Session-list row facts sibling domains read: recency, blank-reuse eligibility, and its cwd canon. */ +export interface SessionsPortSummary { + id: SessionId + /** Empty-log bit (blank sessions are reused by New Session instead of minting another). */ + blank: boolean + cwd?: string + updatedAt: number +} + +/** Session-list facts sibling domains read: readiness, selection, and the row map. */ +export interface SessionsPortList { + ids: SessionId[] + byId: Record + current: SessionId | undefined + phase: 'pending' | 'ready' +} + +/** The sessions-service face injected into sibling domains. */ +export interface SessionsPort { + /** Observable list snapshot (read face only; writes stay inside the sessions domain). */ + readonly list: ObservableSnapshot + /** + * Create a session on the host. + * @param opts - target workspace. + * @returns the new session id. + */ + create(opts: { workspaceId: WorkspaceId }): Promise + /** + * Select a session as current. + * @param id - session id (must exist in the list store). + */ + open(id: SessionId): void + /** Clear the current selection into the no-session view state. */ + clear(): void +} diff --git a/packages/client/runtime/src/client/contract/sessions.ts b/packages/client/runtime/src/client/contract/sessions.ts new file mode 100644 index 0000000000..d26b392072 --- /dev/null +++ b/packages/client/runtime/src/client/contract/sessions.ts @@ -0,0 +1,64 @@ +/** + * The outward sessions-service face — what `ctx.sessions` exposes to feature + * packages and the renderer host, and therefore exactly what the test + * runtime's sessions double must implement. Wire-pump entry points + * (handleMuxEnvelope/handleConnected/refresh) and runtime internals stay on + * the concrete class; cross-domain consumers keep the narrower + * [SessionsPort](./sessions-port.ts). Widening this interface is the + * explicit act of widening what features may do to the sessions domain. + */ +import type { Context } from 'cordis' +import type { SessionId } from '@deepseek-ai/dsh-client-connection/client' +import type { HostObservable, SessionMaybeProvideInfo } from '@deepseek-ai/dsh-client-ui-slots' +import type { + SessionBinding, SessionListState, SessionProvideDescriptor, +} from '../sessions/service.ts' +import type { SessionFace } from './session.ts' +import type { ObservableSnapshot } from './store.ts' + +/** The sessions-service face injected as `ctx.sessions`. */ +export interface ISessions { + /** The useSessions standard feed (list rows + current selection; read face — writes stay inside the domain). */ + readonly list: ObservableSnapshot + /** Atomic current-session provide projection (the renderer host's `sessions.provideInfo` feed). */ + readonly currentProvideInfo: HostObservable + /** + * Select a session as current. + * @param id - session id (must exist in the list; unknown ids fail loud). + */ + open(id: SessionId): void + /** Clear the current selection into the no-session view state. */ + clear(): void + /** + * Register a per-session standard-props provider (hooks become `use` + * selector hooks on the render side; props spread verbatim). + * @param descriptor - static member roster plus per-session resolver. + * @returns disposer removing the provider. + */ + provide(descriptor: SessionProvideDescriptor): () => void + /** + * Resolve an Agent-scoped context view (use-and-discard). + * @param id - session id. + * @returns scoped ctx, or undefined for a session neither listed nor already scoped. + */ + scope(id: SessionId): Context | undefined + /** + * Read the Agent scope tag off a context (service-method seam: fetch + * bundles must reach scope resolution through ctx.sessions). + * @param ctx - any client context. + * @returns the session id, or undefined on root contexts. + */ + scopeOf(ctx: Context): SessionId | undefined + /** + * Resolve the session face behind an Agent-scoped context. + * @param ctx - an Agent-scoped context. + * @returns the session face, or undefined when the ctx is untagged or its scope was pruned. + */ + sessionOf(ctx: Context): SessionFace | undefined + /** + * Resolve the stable session binding (scope-addressed assembly feed). + * @param id - session id. + * @returns binding, or undefined for a session neither listed nor already scoped. + */ + binding(id: SessionId): SessionBinding | undefined +} diff --git a/packages/client/runtime/src/client/contract/workspaces.ts b/packages/client/runtime/src/client/contract/workspaces.ts new file mode 100644 index 0000000000..cd32cbf56f --- /dev/null +++ b/packages/client/runtime/src/client/contract/workspaces.ts @@ -0,0 +1,65 @@ +/** + * The outward workspaces-service face — what `ctx.workspaces` exposes to + * feature packages and the renderer host, and therefore exactly what the + * test runtime's workspaces double must implement. Wire-pump entry points + * (handleHostEnvelope/handleConnected/refresh/startInitialSelection) stay on + * the concrete class. Widening this interface is the explicit act of + * widening what features may do to the workspaces domain. + */ +import type { SessionId, WorkspaceId, WorkspaceView } from '@deepseek-ai/dsh-client-connection/client' +import type { WorkspaceListState } from '../workspaces/service.ts' +import type { ObservableSnapshot } from './store.ts' + +/** The workspaces-service face injected as `ctx.workspaces`. */ +export interface IWorkspaces { + /** The useWorkspaces standard feed (read face — writes stay inside the domain). */ + readonly list: ObservableSnapshot + /** + * Connect a Workspace to its reusable or freshly created blank session. + * @param workspaceId - target workspace. + * @returns the connected session id. + */ + connectWorkspace(workspaceId: WorkspaceId): Promise + /** + * The New Session flow: connect the target (or recent) Workspace and open + * the resulting session; failures surface on the session list state. + * @param workspaceId - explicit target; omitted uses the recency projection. + */ + startSession(workspaceId?: WorkspaceId): void + /** + * Create a Workspace by name or register an existing path. + * @param input - exactly one Host create spelling. + * @returns the created or idempotently resolved Workspace. + */ + create(input: { name: string } | { path: string }): Promise + /** + * Open the Host's native directory picker. + * @returns the selected path, or null when the user cancelled. + */ + pickDirectory(): Promise + /** + * Open a filesystem path with the Host operating system's default application. + * @param path - absolute or host-resolvable path. + */ + openPath(path: string): Promise + /** + * Rename a Workspace. + * @param workspaceId - target workspace. + * @param title - the new display title. + * @returns the updated Workspace view. + */ + rename(workspaceId: WorkspaceId, title: string): Promise + /** + * Delete a Workspace (its sessions fall back to the unaccounted group). + * @param workspaceId - target workspace. + */ + delete(workspaceId: WorkspaceId): Promise + /** + * Move an accounted session within/into a Workspace's ordered list. + * @param workspaceId - target workspace. + * @param sessionId - accounted session to move. + * @param beforeSessionId - accounted anchor to insert before; omitted appends. + * @returns the updated Workspace view. + */ + insertSessionBefore(workspaceId: WorkspaceId, sessionId: SessionId, beforeSessionId?: SessionId): Promise +} diff --git a/packages/client/runtime/src/client/index.ts b/packages/client/runtime/src/client/index.ts index 6cf4318712..780e06f2cd 100644 --- a/packages/client/runtime/src/client/index.ts +++ b/packages/client/runtime/src/client/index.ts @@ -12,10 +12,17 @@ import type { UseProjection } from './sessions/projection-store.ts' export { SlotsService } from './slots.ts' export type { RootOwnerProps } from './slots.ts' export { SessionCreateError, SessionsService, scopeOf, workspaceTitleOf } from './sessions/service.ts' +// The provide channel is shared with the client test runtime (one +// materialization/projection implementation; no test-side mirror to drift). +export { SessionProvideChannel } from './sessions/provide.ts' +export type { SessionProvideChannelHost } from './sessions/provide.ts' export { createScope } from './agents/scope.ts' export type { AgentScopeHandle } from './agents/scope.ts' export { WorkspaceCreateError, WorkspacesService } from './workspaces/service.ts' export type { Session } from './sessions/session.ts' +export type { ISession, ProjectionsFace, SessionFace } from './contract/session.ts' +export type { ISessions } from './contract/sessions.ts' +export type { IWorkspaces } from './contract/workspaces.ts' export type { SessionBinding, SessionListState, SessionProvideContribution, SessionProvideDescriptor, SessionSummary, } from './sessions/service.ts' @@ -118,8 +125,10 @@ declare module 'cordis' { } interface Context { slots: import('./slots.ts').SlotsService - sessions: import('./sessions/service.ts').SessionsService - workspaces: import('./workspaces/service.ts').WorkspacesService + /** The outward face only; the concrete service stays inside the runtime. */ + sessions: import('./contract/sessions.ts').ISessions + /** The outward face only; the concrete service stays inside the runtime. */ + workspaces: import('./contract/workspaces.ts').IWorkspaces } } diff --git a/packages/client/runtime/src/client/sessions/provide.ts b/packages/client/runtime/src/client/sessions/provide.ts new file mode 100644 index 0000000000..8039d7faf7 --- /dev/null +++ b/packages/client/runtime/src/client/sessions/provide.ts @@ -0,0 +1,190 @@ +/** + * The session standard-props provide channel: provider roster, bundle + * materialization (fail-loud on undeclared/missing/duplicate members), the + * static no-session projection, and the atomic current-session projection + * observable. One implementation — SessionsService drives it from wire + * truth, the test runtime's sessions double drives it from fixtures — so + * the materialization rules and the projection semantics cannot drift + * between production and the test bench. + */ +import type { HostObservable, SessionMaybeProvideInfo, SessionProvideInfo } from '@deepseek-ai/dsh-client-ui-slots' +import type { SessionBinding, SessionProvideDescriptor } from './service.ts' + +/** The owner-side hooks: how the channel reaches the owner's live bundles and current selection. */ +export interface SessionProvideChannelHost { + /** + * Re-materialize every already-materialized bundle against the new roster + * (call {@link SessionProvideChannel.materializeInfo} per live binding). + * Lazily-materialized sessions pick the new roster up on first resolve. + */ + rebuildBundles(): void + /** Resolve the current selection's bundle (the owner's maybe-provide lookup). */ + resolveCurrent(): SessionMaybeProvideInfo +} + +/** + * Provider roster + materialization + current projection. The channel owns + * every rule a provider contribution must satisfy; owners keep only their + * per-session bundle storage and the definition of "current". + */ +export class SessionProvideChannel { + private readonly providers: SessionProvideDescriptor[] = [] + private maybeInfoCache: SessionMaybeProvideInfo + /** Latest published current bundle (identity comparison dedupes republish). */ + private currentSnapshot: SessionMaybeProvideInfo + /** Projection subscribers (plain cell: bundles hold live session sources, so no store freeze may touch them). */ + private readonly listeners = new Set<() => void>() + + /** + * Atomic current-session provide projection: selection changes and + * provider-roster changes publish through this one source, so a roster + * change under a stable current id republishes the bundle instead of + * stranding mounted entries. + */ + readonly currentProvideInfo: HostObservable + + /** + * @param host - owner-side bundle storage and current-selection resolution. + */ + constructor(private readonly host: SessionProvideChannelHost) { + // The runtime's own contribution comes first: useSession rides the same + // provide channel every plugin uses (no renderer special case). + this.providers.push({ + hooks: ['session'], + resolve: binding => ({ hooks: { session: binding.session } }), + }) + this.maybeInfoCache = this.materializeMaybeInfo() + this.currentSnapshot = this.maybeInfoCache + this.currentProvideInfo = { + getSnapshot: () => this.currentSnapshot, + subscribe: (fn) => { + this.listeners.add(fn) + return () => { this.listeners.delete(fn) } + }, + } + } + + /** The static no-session projection under the current roster (declared names present, values undefined). */ + get maybeInfo(): SessionMaybeProvideInfo { + return this.maybeInfoCache + } + + /** + * Register a per-session standard-props provider (see + * SessionsService.provide for the product contract). Live bundles rebuild + * immediately; misdeclared providers fail loud here, at the registration + * edge, and the registration rolls back — the channel never stays on a + * roster it cannot materialize. + * @param descriptor - static member roster plus per-session resolver. + * @returns disposer removing the provider. + */ + provide(descriptor: SessionProvideDescriptor): () => void { + this.providers.push(descriptor) + try { + this.applyRosterChange() + } catch (error) { + this.providers.splice(this.providers.indexOf(descriptor), 1) + // Restore the previous (valid) roster's bundles; cannot rethrow — the + // pre-push roster materialized successfully before. + this.applyRosterChange() + throw error + } + return () => { + const at = this.providers.indexOf(descriptor) + if (at >= 0) this.providers.splice(at, 1) + this.applyRosterChange() + } + } + + /** + * Re-derive the current selection's bundle and publish it when it changed. + * Bundles are identity-stable per (scope, roster) materialization, so an + * identity compare is exact; synchronous notify — call sites (the owner's + * list subscription, provide()) already sit behind their own batching or + * registration edges. + */ + publishCurrent(): void { + const next = this.host.resolveCurrent() + if (next === this.currentSnapshot) return + this.currentSnapshot = next + for (const fn of [...this.listeners]) { + try { + fn() + } catch (error) { + // Contain subscriber failures: this notify runs inside the list + // notification, where a throwing render-side subscriber would starve + // later listeners and abort the projection pass that scheduled it. + console.error('sessions.currentProvideInfo subscriber failed:', error) + } + } + } + + /** + * Materialize the standard-props bundle for one session (fails loud on + * undeclared, missing, and duplicate member names). + * @param binding - session assembly handle fed to every resolver. + * @returns the materialized bundle (identity-stable until the next materialization). + */ + materializeInfo(binding: SessionBinding): SessionProvideInfo { + const hooks: Record> = {} + const props: Record = {} + for (const descriptor of this.providers) { + const contribution = descriptor.resolve(binding) + const contributedHooks = contribution.hooks ?? {} + const contributedProps = contribution.props ?? {} + for (const name of Object.keys(contributedHooks)) { + if (!(descriptor.hooks ?? []).includes(name)) { + throw new Error(`sessions.provide: undeclared hook "${name}"`) + } + } + for (const name of Object.keys(contributedProps)) { + if (!(descriptor.props ?? []).includes(name)) { + throw new Error(`sessions.provide: undeclared prop "${name}"`) + } + } + for (const name of descriptor.hooks ?? []) { + const source = contributedHooks[name] + if (source === undefined) throw new Error(`sessions.provide: missing hook "${name}"`) + if (Object.hasOwn(hooks, name)) throw new Error(`sessions.provide: duplicate hook "${name}"`) + hooks[name] = source + } + for (const name of descriptor.props ?? []) { + if (!Object.hasOwn(contributedProps, name)) throw new Error(`sessions.provide: missing prop "${name}"`) + if (Object.hasOwn(props, name)) throw new Error(`sessions.provide: duplicate prop "${name}"`) + props[name] = contributedProps[name] + } + } + return { + sessionId: binding.sessionId, + hooks, + props, + // The useProjection seat: key-addressed bare value faces off the + // session's projection store (open key space — never a static roster member). + projections: { faceOf: key => binding.session.projections.faceOf(key) }, + } + } + + /** Rebuild the static projection and the owner's live bundles, then republish the current one. */ + private applyRosterChange(): void { + this.maybeInfoCache = this.materializeMaybeInfo() + this.host.rebuildBundles() + this.publishCurrent() + } + + /** Build the static no-session kit and reject duplicate declared names. */ + private materializeMaybeInfo(): SessionMaybeProvideInfo { + const hooks: Record = {} + const props: Record = {} + for (const descriptor of this.providers) { + for (const name of descriptor.hooks ?? []) { + if (Object.hasOwn(hooks, name)) throw new Error(`sessions.provide: duplicate hook "${name}"`) + hooks[name] = undefined + } + for (const name of descriptor.props ?? []) { + if (Object.hasOwn(props, name)) throw new Error(`sessions.provide: duplicate prop "${name}"`) + props[name] = undefined + } + } + return { sessionId: undefined, hooks, props } // no projections face: every key reads absent without a session + } +} diff --git a/packages/client/runtime/src/client/sessions/service.ts b/packages/client/runtime/src/client/sessions/service.ts index 19e022d450..64680587e2 100644 --- a/packages/client/runtime/src/client/sessions/service.ts +++ b/packages/client/runtime/src/client/sessions/service.ts @@ -22,9 +22,12 @@ import type { } from '@deepseek-ai/dsh-client-ui-slots' import type { SnapshotStore } from '../contract/store.ts' import { createSnapshotStore } from '../contract/store.ts' +import type { SessionFace } from '../contract/session.ts' +import type { ISessions } from '../contract/sessions.ts' import { createScope, scopeOf as scopeTagOf } from '../agents/scope.ts' import { SessionManager } from './manager.ts' import type { SessionListPhase } from './manager.ts' +import { SessionProvideChannel } from './provide.ts' import type { Session } from './session.ts' /** Session list row projected from the host list RPC plus live stream increments. */ @@ -79,7 +82,8 @@ export class SessionCreateError extends Error { /** Session assembly handle for SessionProvider/inject factories (identity-stable per session). */ export interface SessionBinding { readonly sessionId: SessionId - readonly session: Session + /** The outward session face only — feature code never sees the concrete class. */ + readonly session: SessionFace readonly ctx: Context } @@ -119,6 +123,8 @@ interface ScopeRecord { fiber: Fiber ctx: Context binding: SessionBinding + /** The concrete Session for runtime-internal entry points (staging open()); the binding carries only the outward face. */ + session: Session /** Render-layer standard-props bundle (identity-stable per scope; the renderer's per-info caches key off it). */ provideInfo: SessionProvideInfo } @@ -146,7 +152,7 @@ export interface SessionProvideDescriptor { } /** Root sessions service: list store, current selection, object-layer manager, scope tree, bindings, ancestry. */ -export class SessionsService { +export class SessionsService implements ISessions { /** List snapshot store (list RPC + host stream increments; re-pulled on reconnect) — the useSessions standard feed, current included. */ readonly list: SnapshotStore /** The object-layer instance cluster and frame dispatch entry. */ @@ -170,14 +176,8 @@ export class SessionsService { private readonly selection: SnapshotStore<{ sessionId?: SessionId }> private readonly scopes = new Map() - /** Registered per-session standard-props providers, in registration order. */ - private readonly providers: SessionProvideDescriptor[] = [] - /** Static no-session projection, rebuilt only when the provider roster changes. */ - private maybeInfo: SessionMaybeProvideInfo - /** Latest published {@link SessionsService.currentProvideInfo} bundle (identity comparison dedupes republish). */ - private currentProvideInfoSnapshot: SessionMaybeProvideInfo - /** currentProvideInfo subscribers (plain cell: bundles hold live Session sources, so no store freeze may touch them). */ - private readonly currentProvideInfoListeners = new Set<() => void>() + /** The provide channel (roster, materialization rules, current projection) — shared with the test runtime's double. */ + private readonly provideChannel: SessionProvideChannel /** * The staged session id — follows `list.current` exactly, holding its last * defined value across masked gaps (a transiently absent selection blanks @@ -212,23 +212,17 @@ export class SessionsService { // The current-provide projection follows the same current writes. this.list.subscribe(() => { this.followCurrent() - this.updateCurrentProvideInfo() + this.provideChannel.publishCurrent() }) - // The runtime's own contribution comes first: useSession rides the same - // provide channel every plugin uses (no renderer special case). - this.providers.push({ - hooks: ['session'], - resolve: binding => ({ hooks: { session: binding.session } }), - }) - this.maybeInfo = this.materializeMaybeProvideInfo() - this.currentProvideInfoSnapshot = this.maybeInfo - this.currentProvideInfo = { - getSnapshot: () => this.currentProvideInfoSnapshot, - subscribe: (fn) => { - this.currentProvideInfoListeners.add(fn) - return () => { this.currentProvideInfoListeners.delete(fn) } + this.provideChannel = new SessionProvideChannel({ + rebuildBundles: () => { + for (const record of this.scopes.values()) { + record.provideInfo = this.provideChannel.materializeInfo(record.binding) + } }, - } + resolveCurrent: () => this.maybeProvideInfo(this.list.getSnapshot().current), + }) + this.currentProvideInfo = this.provideChannel.currentProvideInfo rootCtx.reflect.provide('sessions', this, undefined) } @@ -243,105 +237,10 @@ export class SessionsService { * @returns disposer removing the provider (already-materialized bundles keep their members until their scope drops). */ provide(descriptor: SessionProvideDescriptor): () => void { - this.providers.push(descriptor) // Scopes may already exist (boot order: the list lands and resolves - // scopes before later plugins register) — their bundles must include - // every provider by first render, so re-materialize on roster change. - this.rematerializeProvideBundles() - return () => { - const at = this.providers.indexOf(descriptor) - if (at >= 0) this.providers.splice(at, 1) - this.rematerializeProvideBundles() - } - } - - /** Rebuild every live scope's standard-props bundle after a provider roster change. */ - private rematerializeProvideBundles(): void { - this.maybeInfo = this.materializeMaybeProvideInfo() - for (const record of this.scopes.values()) { - record.provideInfo = this.materializeProvideInfo(record.binding) - } - this.updateCurrentProvideInfo() - } - - /** - * Re-derive the current selection's provide bundle and publish it when it - * changed. Bundles are identity-stable per (scope, roster) - * materialization, so an identity compare is exact; synchronous notify — - * both call sites (list.subscribe, provide()) already sit behind their own - * batching or registration edges. - */ - private updateCurrentProvideInfo(): void { - const next = this.maybeProvideInfo(this.list.getSnapshot().current) - if (next === this.currentProvideInfoSnapshot) return - this.currentProvideInfoSnapshot = next - for (const fn of [...this.currentProvideInfoListeners]) { - try { - fn() - } catch (error) { - // Contain subscriber failures: this notify runs inside the list - // notification, where a throwing render-side subscriber would starve - // later listeners and abort the projection pass that scheduled it. - console.error('sessions.currentProvideInfo subscriber failed:', error) - } - } - } - - /** Build the static no-session kit and reject duplicate declared names. */ - private materializeMaybeProvideInfo(): SessionMaybeProvideInfo { - const hooks: Record = {} - const props: Record = {} - for (const descriptor of this.providers) { - for (const name of descriptor.hooks ?? []) { - if (Object.hasOwn(hooks, name)) throw new Error(`sessions.provide: duplicate hook "${name}"`) - hooks[name] = undefined - } - for (const name of descriptor.props ?? []) { - if (Object.hasOwn(props, name)) throw new Error(`sessions.provide: duplicate prop "${name}"`) - props[name] = undefined - } - } - return { sessionId: undefined, hooks, props } // no projections face: every key reads absent without a session - } - - /** Materialize the standard-props bundle for one session (fails loud on duplicate member names). */ - private materializeProvideInfo(binding: SessionBinding): SessionProvideInfo { - const hooks: Record> = {} - const props: Record = {} - for (const descriptor of this.providers) { - const contribution = descriptor.resolve(binding) - const contributedHooks = contribution.hooks ?? {} - const contributedProps = contribution.props ?? {} - for (const name of Object.keys(contributedHooks)) { - if (!(descriptor.hooks ?? []).includes(name)) { - throw new Error(`sessions.provide: undeclared hook "${name}"`) - } - } - for (const name of Object.keys(contributedProps)) { - if (!(descriptor.props ?? []).includes(name)) { - throw new Error(`sessions.provide: undeclared prop "${name}"`) - } - } - for (const name of descriptor.hooks ?? []) { - const source = contributedHooks[name] - if (source === undefined) throw new Error(`sessions.provide: missing hook "${name}"`) - if (Object.hasOwn(hooks, name)) throw new Error(`sessions.provide: duplicate hook "${name}"`) - hooks[name] = source - } - for (const name of descriptor.props ?? []) { - if (!Object.hasOwn(contributedProps, name)) throw new Error(`sessions.provide: missing prop "${name}"`) - if (Object.hasOwn(props, name)) throw new Error(`sessions.provide: duplicate prop "${name}"`) - props[name] = contributedProps[name] - } - } - return { - sessionId: binding.sessionId, - hooks, - props, - // The useProjection seat: key-addressed bare value faces off the - // session's projection store (open key space — never a static roster member). - projections: { faceOf: key => binding.session.projections.faceOf(key) }, - } + // scopes before later plugins register) — the channel rebuilds their + // bundles through the host hooks so every provider lands by first render. + return this.provideChannel.provide(descriptor) } /** @@ -439,9 +338,9 @@ export class SessionsService { * `agent.session`). Same service-method seam as * {@link SessionsService.scopeOf}. * @param ctx - an Agent-scoped context. - * @returns the Session, or undefined when the ctx is untagged or its scope was pruned. + * @returns the session face, or undefined when the ctx is untagged or its scope was pruned. */ - sessionOf(ctx: Context): Session | undefined { + sessionOf(ctx: Context): SessionFace | undefined { const id = scopeTagOf(ctx) if (id === undefined) return undefined return this.scopes.get(id)?.binding.session @@ -473,7 +372,7 @@ export class SessionsService { * return the static no-session projection rather than removing hook props. */ private maybeProvideInfo(id: string | undefined): SessionMaybeProvideInfo { - return (id === undefined ? undefined : this.provideInfo(id)) ?? this.maybeInfo + return (id === undefined ? undefined : this.provideInfo(id)) ?? this.provideChannel.maybeInfo } /** @@ -497,7 +396,7 @@ export class SessionsService { * validates and the projection masks absent selections), so resolve * cannot miss; kept so a future current writer cannot crash the notify. */ if (record !== undefined) { - void record.binding.session.open() + void record.session.open() } } @@ -540,8 +439,9 @@ export class SessionsService { fiber, ctx, binding, + session, // Sources are bare observables; React binds selector hooks at its own seam. - provideInfo: this.materializeProvideInfo(binding), + provideInfo: this.provideChannel.materializeInfo(binding), } this.scopes.set(id, record) return record @@ -608,7 +508,7 @@ export class SessionsService { void record.fiber.dispose() // Release the Session's dispatch point with the scope it belongs to (a // surviving instance — the live Intent — rebinds when resolve re-mints). - record.binding.session.unbindScope() + record.session.unbindScope() // Optional lookup: slots and sessions are sibling services with no // declared dependency; a slots-less boot (object-layer tests) skips. this.rootCtx.get('slots')?.pruneStoreScope(id) diff --git a/packages/client/runtime/src/client/sessions/session.ts b/packages/client/runtime/src/client/sessions/session.ts index 2fedaa9227..6f58589cb6 100644 --- a/packages/client/runtime/src/client/sessions/session.ts +++ b/packages/client/runtime/src/client/sessions/session.ts @@ -10,7 +10,7 @@ import type { // Value import from the inline-safe wire layer (not the connection plugin): // plugin-to-plugin value imports are a bundle purity error. import { transportError } from '@deepseek-ai/dsh-host-apiproxy/api' -import type { ObservableSnapshot } from '../contract/store.ts' +import type { SessionFace } from '../contract/session.ts' import type { CodeSubCall, ComposerPhase, ConversationNode, ConversationSnapshot, OpenState, PromptError, QueuedMessage, RunningToolCall, @@ -73,11 +73,13 @@ function queuePreviewOf(content: readonly ContentBlock[]): string { } /** - * Owns a session's event window and exposes two observable read surfaces: - * the folded chat conversation and the raw history window. React bindings - * remain outside this data layer. + * Owns a session's event window, folded conversation, raw history inspection, + * and observable snapshot. React bindings remain outside this data layer. + * Features see only the {@link SessionFace} slice (ISession verbs + the + * snapshot source); the remaining public members are manager/runtime entry + * points. */ -export class Session implements ObservableSnapshot { +export class Session implements SessionFace { // ---- Window and derived state (all private; the snapshot is the only read surface) ---- private events: SessionEvent[] = [] /** Wire views aligned with `events` by index (envelope-level annotations; undefined = no view). @@ -231,12 +233,14 @@ export class Session implements ObservableSnapshot { this.notifier.markDirty() return result } - // Blank flips on ACCEPTANCE, not attempt: an accepted prompt has logged - // its user/message on the host (events.length > 0 is fact, not - // optimism), while a rejected first prompt must keep the session blank - // — the client-side blank mirror only ever lowers, so flipping early on - // a failure would surface the session forever and strip its - // connectWorkspace reuse eligibility against the host's authority. + // Blank flips on ACCEPTANCE, not attempt: an accepted prompt starts the + // conversation's first turn on the host (the host criterion — a logged + // turn/start — is fact, not optimism; standalone command and projection + // events never flip it), while a rejected first prompt must keep the + // session blank — the client-side blank mirror only ever lowers, so + // flipping early on a failure would surface the session forever and + // strip its connectWorkspace reuse eligibility against the host's + // authority. if (this.blankBit) { this.blankBit = false this.options.onEngaged?.(this) diff --git a/packages/client/runtime/src/client/workspaces/service.ts b/packages/client/runtime/src/client/workspaces/service.ts index 97f01d0bf1..c69661b130 100644 --- a/packages/client/runtime/src/client/workspaces/service.ts +++ b/packages/client/runtime/src/client/workspaces/service.ts @@ -6,7 +6,8 @@ import type { } from '@deepseek-ai/dsh-client-connection/client' import type { SnapshotStore } from '../contract/store.ts' import { createSnapshotStore } from '../contract/store.ts' -import type { SessionsService } from '../sessions/service.ts' +import type { SessionsPort, SessionsPortList } from '../contract/sessions-port.ts' +import type { IWorkspaces } from '../contract/workspaces.ts' import { WorkspaceManager, type WorkspaceListPhase } from './manager.ts' /** Workspace list plus the two-baseline readiness and default-target projection. */ @@ -30,7 +31,7 @@ export class WorkspaceCreateError extends Error { } /** Real Workspace object layer and Host actions. */ -export class WorkspacesService { +export class WorkspacesService implements IWorkspaces { /** UI-facing immutable projection; the manager remains wire truth. */ readonly list: SnapshotStore /** Workspace baseline and frame owner. */ @@ -43,9 +44,9 @@ export class WorkspacesService { /** * @param ctx - client root context. * @param api - shared wire client. - * @param sessions - lower-level Session service used for recency and blank-session reuse. + * @param sessions - cross-domain sessions face used for recency and blank-session reuse. */ - constructor(ctx: Context, private readonly api: IApiClient, private readonly sessions: SessionsService) { + constructor(ctx: Context, private readonly api: IApiClient, private readonly sessions: SessionsPort) { this.manager = new WorkspaceManager(api) this.list = createSnapshotStore({ items: [], state: 'idle', phase: 'pending', error: null, @@ -271,7 +272,7 @@ export class WorkspacesService { /** Stable tie-breaking follows Host Workspace order. */ function recentWorkspace( workspaces: readonly WorkspaceView[], - sessions: ReturnType['byId'], + sessions: SessionsPortList['byId'], ): WorkspaceId | undefined { let selected: WorkspaceId | undefined let selectedTime = Number.NEGATIVE_INFINITY diff --git a/packages/client/runtime/tests/fake-api.ts b/packages/client/runtime/tests/fake-api.ts index b18ae87d8e..bf1a4a04b3 100644 --- a/packages/client/runtime/tests/fake-api.ts +++ b/packages/client/runtime/tests/fake-api.ts @@ -153,6 +153,15 @@ export class FakeApiClient implements IApiClient { list: (payload: unknown) => this.record('skill.list', payload, this.onSkillList(payload)), } + readonly goals: IApiClient['goals'] = { + create: payload => this.record('goal.create', payload, Promise.resolve(ok({ ref: { id: 'fake-goal' as never, revision: 1 } }))), + edit: payload => this.record('goal.edit', payload, Promise.resolve(ok({ ref: { id: 'fake-goal' as never, revision: 1 } }))), + pause: payload => this.record('goal.pause', payload, Promise.resolve(ok({ ref: { id: 'fake-goal' as never, revision: 1 } }))), + resume: payload => this.record('goal.resume', payload, Promise.resolve(ok({ ref: { id: 'fake-goal' as never, revision: 1 } }))), + complete: payload => this.record('goal.complete', payload, Promise.resolve(ok({ ref: { id: 'fake-goal' as never, revision: 1 } }))), + clear: payload => this.record('goal.clear', payload, Promise.resolve(ok({ cleared: true as const }))), + } + /** When true, streams never fire onOpen (misbehaving-carrier material for the handshake timeout guard). */ suppressStreamOpen = false diff --git a/packages/client/test-runtime/README.i18n.yaml b/packages/client/test-runtime/README.i18n.yaml new file mode 100644 index 0000000000..73a4705b1b --- /dev/null +++ b/packages/client/test-runtime/README.i18n.yaml @@ -0,0 +1,6 @@ +# Bilingual-pair consistency record (docs/i18n/README.md): the git blob hash of each +# side as of the last confirmed-consistent state. Both languages carry equal authority; +# after editing either side, bring the other along and re-record with: +# pnpm run verify-translation-pairing --write packages/client/test-runtime/README.md +README.md: 883d71224139dc409229fdfb35362d040e810cc7 +README.zh.md: a3daf112940b03b585d44bc5fd1317e43ebd35fe diff --git a/packages/client/test-runtime/README.md b/packages/client/test-runtime/README.md new file mode 100644 index 0000000000..883d712241 --- /dev/null +++ b/packages/client/test-runtime/README.md @@ -0,0 +1,24 @@ +# @deepseek-ai/dsh-client-test-runtime + +English | [中文](README.zh.md) + +jsdom slot test runtime for client feature specs: a real Cordis `Context`, the production `SlotsService` and web-react renderer, assembled around typed session/workspace doubles. Feature suites exercise declaration, registration, scope, store, inject, rendering, updates, and disposal without hand-building the machinery per suite — and without a second implementation of any production logic. + +The doubles implement the same outward faces features receive through ctx (`TestSessions implements ISessions`, `TestWorkspaces implements IWorkspaces`; each fixture session is a `FixtureSession implements SessionFace`), so a production face change breaks the bench at compile time instead of silently drifting. Provide-bundle materialization runs the production `SessionProvideChannel` — the one implementation shared with `SessionsService`. Fixtures feed plain data: list rows, conversation snapshots (immer-patched via `updateSnapshot`), projection values, and `ISession`-typed behavior stubs that fail loud when a spec calls an unstubbed verb. The typed `provide()` constrains fakes for declared service names to `Partial` of that service's outward face. + +Local DOM snapshots: `declare(children)` registers an auto frame whose per-key `
` wrappers are snapshot roots; `renderSlot(key, owner)` returns the slot-local view (container, scoped Testing Library queries, in-place `update(owner)`); a registered snapshot serializer folds CSS-module class hashes (`_frame_a1b2c3` → `frame`) to keep `.snap` files structural and collapses `` internals to a `data-content` fingerprint. Suites needing a custom page frame use `root.declare(children, Frame)` instead; `mount(plugin)` runs a real fiber with fail-loud service prechecks, and `dispose()` tears down views, feature fibers, minted scopes, and persisted store state on one axis. + +Not part of the product plugin graph (no `dshClient`); feature packages depend on it in `devDependencies` only. + +## Model Experience + +None, as this package is browser-side test infrastructure; nothing here reaches a model request. + +#### KV Cache effect + +None; this package neither assembles nor sends a provider request. + +## Known Limitations and Deferred Work + +- **Consumed through repository source aliases only.** Specs resolve the package through tsconfig `paths` to `src`; the built `lib/` artifact re-exports `@deepseek-ai/dsh-client-runtime/client`, whose bundle is a browser loader script with no Node ESM exports, so `lib/index.js` is not importable under plain Node. Acceptable while every consumer is an in-repo Vitest suite; a Node-compatible runtime entry is deferred until an out-of-repo consumer exists. +- **Conversation snapshots are fixture data, not replayed history.** `updateSnapshot` writes the snapshot store directly; the wire-to-snapshot computation stays covered by the runtime package's own tests and the replay e2e. A fixture can therefore express states the production fold would never produce. diff --git a/packages/client/test-runtime/README.zh.md b/packages/client/test-runtime/README.zh.md new file mode 100644 index 0000000000..a3daf11294 --- /dev/null +++ b/packages/client/test-runtime/README.zh.md @@ -0,0 +1,24 @@ +# @deepseek-ai/dsh-client-test-runtime + +[English](README.md) | 中文 + +面向 client feature 测试的 jsdom slot 测试运行时：真实 Cordis `Context`、生产 `SlotsService` 与 web-react 渲染器，围绕带类型的 session/workspace 测试替身组装。feature 套件无需逐套件手搭机器即可测遍声明、注册、scope、store、inject、渲染、更新与销毁——且不存在任何生产逻辑的第二份实现。 + +替身实现的正是 feature 经 ctx 拿到的对外面（`TestSessions implements ISessions`、`TestWorkspaces implements IWorkspaces`；每个 fixture session 是 `FixtureSession implements SessionFace`），生产面一旦改形，测试台在编译期即断，而非静默漂移。provide bundle 材料化直接运行生产 `SessionProvideChannel`——与 `SessionsService` 共用同一份实现。fixture 灌入的是普通数据：列表行、会话快照（经 `updateSnapshot` 以 immer 补丁改写）、projection 值，以及按 `ISession` 取型的行为桩——spec 调用未打桩的动词时报错自明。带类型的 `provide()` 将已声明服务名的 fake 约束为该服务对外面的 `Partial` 子集。 + +局部 DOM 快照：`declare(children)` 注册自动 frame，逐 key 的 `
` 包裹层即快照根；`renderSlot(key, owner)` 返回该 slot 的局部视图（container、限定范围的 Testing Library 查询、原位 `update(owner)`）；注册的快照序列化器把 CSS-module 哈希类名折回语义名（`_frame_a1b2c3` → `frame`）保持 `.snap` 只含结构，并把 `` 内部折叠为 `data-content` 指纹。需要自定义页面 frame 的套件改用 `root.declare(children, Frame)`；`mount(plugin)` 在真实 fiber 上运行并对缺失服务先行报错；`dispose()` 沿单一轴拆除视图、feature fiber、已铸 scope 与持久化 store 状态。 + +不属于产品插件图（无 `dshClient`）；feature 包仅以 `devDependencies` 依赖之。 + +## Model Experience + +无；本包是浏览器侧测试基础设施，无一物到达模型请求。 + +#### KV Cache effect + +无；本包既不组装也不发送 provider 请求。 + +## Known Limitations and Deferred Work + +- **仅可经仓内源码别名消费。** spec 通过 tsconfig `paths` 解析到 `src`；构建产物 `lib/` 再导出 `@deepseek-ai/dsh-client-runtime/client`，而该 bundle 是无 Node ESM 导出的浏览器 loader 脚本，故 `lib/index.js` 在纯 Node 下不可导入。当前所有消费方都是仓内 Vitest 套件，可接受；Node 兼容的运行时入口待出现仓外消费方再补。 +- **会话快照是 fixture 数据，不是重放历史。** `updateSnapshot` 直写快照 store；wire 到快照的运算仍由 runtime 包自身测试与 replay e2e 把守。因此 fixture 可以表达生产折叠永不产出的状态。 diff --git a/packages/client/test-runtime/package.json b/packages/client/test-runtime/package.json new file mode 100644 index 0000000000..10b80bdd68 --- /dev/null +++ b/packages/client/test-runtime/package.json @@ -0,0 +1,54 @@ +{ + "name": "@deepseek-ai/dsh-client-test-runtime", + "description": "jsdom slot test runtime: real Cordis Context + SlotsService + web-react renderer with test-owned session/workspace doubles for feature specs", + "version": "0.0.1", + "private": true, + "type": "module", + "main": "lib/index.js", + "types": "lib/types/index.d.ts", + "exports": { + ".": { + "types": "./lib/types/index.d.ts", + "default": "./lib/index.js" + }, + "./invariant": { + "types": "./lib/types/invariant.d.ts", + "default": "./lib/invariant.js" + }, + "./src/*": "./src/*", + "./package.json": "./package.json" + }, + "license": "BSD-3-Clause", + "dependencies": { + "@testing-library/dom": "^10.4.1", + "@testing-library/react": "^16.3.2", + "vitest": "^4.1.8" + }, + "peerDependencies": { + "@deepseek-ai/dsh-client-runtime": "^0.0.1", + "@deepseek-ai/dsh-client-ui-slots": "^0.0.1", + "@deepseek-ai/dsh-client-web-react": "^0.0.1", + "@deepseek-ai/dsh-invariants": "^0.0.1", + "cordis": "^4.0.0-rc.7", + "react": "^18.2.0", + "react-dom": "^18.2.0" + }, + "devDependencies": { + "@deepseek-ai/dsh-client-runtime": "workspace:^", + "@deepseek-ai/dsh-client-ui-slots": "workspace:^", + "@deepseek-ai/dsh-client-web-react": "workspace:^", + "@deepseek-ai/dsh-invariants": "workspace:^", + "@types/react": "~18.3.1", + "@types/react-dom": "~18.3.0", + "cordis": "^4.0.0-rc.7", + "react": "^18.2.0", + "react-dom": "^18.2.0" + }, + "files": [ + "lib/index.js", + "lib/invariant.js", + "lib/types/**/*.d.ts", + "lib/types/**/*.d.ts.map", + "src" + ] +} diff --git a/packages/client/test-runtime/src/fixtures.ts b/packages/client/test-runtime/src/fixtures.ts new file mode 100644 index 0000000000..4219d233e2 --- /dev/null +++ b/packages/client/test-runtime/src/fixtures.ts @@ -0,0 +1,82 @@ +/** Session/workspace fixture shapes and snapshot defaults for the test runtime. */ +import type { + ConversationSnapshot, ISession, SessionId, SessionSummary, WorkspaceListState, +} from '@deepseek-ai/dsh-client-runtime/client' + +/** + * Fixture overrides for the session behavior face: any subset of the + * production ISession verbs (typed against it, so a face change surfaces + * here at compile time), plus extra members feature-specific casts consume. + * The open Record tail means a misnamed EXTRA member is not caught by the + * compiler (it grafts as dead weight); the ISession verbs stay safe — a + * misnamed verb leaves the fail-loud stub in place, which names itself at + * the first call. + */ +export type SessionBehaviorOverrides = Partial & Record + +/** + * act-wrapped mutation runner shared by every runtime object: public mutators + * funnel through it so tests never handle SlotCore microtask batching or + * React act themselves. + */ +export type Stabilizer = (fn: () => void | Promise) => Promise + +/** + * Session fixture accepted by {@link TestSessions.add}: identity plus optional + * snapshot/list-row overrides and the session behavior face the feature under + * test actually calls (kept open — the runtime never fakes methods a test did + * not supply, so an unstubbed call fails loud at the call site). + */ +export interface SessionFixture { + id: string + /** Overrides merged over {@link conversationSnapshot} (sessionId comes from `id`). */ + snapshot?: Partial> + /** List-row overrides merged over the defaults derived from `id`. */ + summary?: Partial> + /** Session behavior face: exactly the methods the feature under test calls (ISession subset + extras). */ + session?: SessionBehaviorOverrides +} + +/** + * A complete quiescent conversation snapshot (open window, no traffic). + * @param sessionId - owning session id. + * @returns the snapshot; spread fixture overrides on top. + */ +export function conversationSnapshot(sessionId: SessionId): ConversationSnapshot { + return { + sessionId, + nodes: [], + foldDegraded: false, + partial: null, + runningCalls: [], + codeDispatches: new Map(), + pending: [], + queue: [], + running: false, + composerPhase: 'active', + removed: false, + openState: 'open', + openError: null, + hasMore: false, + loadingOlder: false, + promptError: null, + blank: false, + lastAgentError: null, + } +} + +/** + * A ready workspace list with no workspaces (the shape WorkspacesService + * projects after both baselines land). + * @returns the initial state of the test workspaces store. + */ +export function workspaceListState(): WorkspaceListState { + return { + items: [], + state: 'idle', + phase: 'ready', + error: null, + baselinesReady: true, + recentWorkspaceId: undefined, + } +} diff --git a/packages/client/test-runtime/src/index.ts b/packages/client/test-runtime/src/index.ts new file mode 100644 index 0000000000..1b70e10d79 --- /dev/null +++ b/packages/client/test-runtime/src/index.ts @@ -0,0 +1,372 @@ +/** + * jsdom slot test runtime: a real small runtime — Cordis `Context`, the + * runtime `SlotsService`, and the web-react renderer — assembled around + * test-owned session/workspace doubles, so feature specs exercise + * declaration, registration, scope, store, inject, rendering, updates, and + * disposal without hand-building the machinery per suite. + * + * Not part of the product plugin graph (no `dshClient`); feature packages + * depend on it in devDependencies only. It copies no SlotCore/renderer/store + * machinery — everything mounts the production implementations. + * @module @deepseek-ai/dsh-client-test-runtime + */ +/* eslint-disable @typescript-eslint/no-redundant-type-constituents -- + * `keyof SlotMap & string` is the declare-merge key pattern (see ui-slots): + * this compilation unit sees only the runtime's 'root' row, but consumer + * programs merge their own keys in; the rule fires on the narrow-map view. */ +import { Context, Inject } from 'cordis' +import type { Fiber, Plugin } from 'cordis' +import { createElement, Fragment, useSyncExternalStore } from 'react' +import type { ReactNode } from 'react' +import { act, render, within } from '@testing-library/react' +import type { RenderResult } from '@testing-library/react' +import type { queries } from '@testing-library/dom' +import type { BoundFunctions } from '@testing-library/dom' +import { SlotsService } from '@deepseek-ai/dsh-client-runtime/client' +import { createSlotRenderer } from '@deepseek-ai/dsh-client-web-react' +import type { + ChildrenDecl, ComposedProps, OwnerOf, SlotComponent, SlotMap, SlotRendererHost, StoreInstanceLike, +} from '@deepseek-ai/dsh-client-ui-slots' +import { registerDomSnapshotSerializer } from './snapshot.ts' +import { TestSessions } from './sessions.ts' +import { TestWorkspaces } from './workspaces.ts' +import type { Stabilizer } from './fixtures.ts' + +export { domSnapshotSerializer, registerDomSnapshotSerializer } from './snapshot.ts' +export { FixtureSession, TestSessions } from './sessions.ts' +export { TestWorkspaces } from './workspaces.ts' +export { conversationSnapshot, workspaceListState } from './fixtures.ts' +export type { SessionBehaviorOverrides, SessionFixture, Stabilizer } from './fixtures.ts' + +/** Erased register face for the internal root call (the public declare seam holds the typing). */ +type ErasedRegister = (options: object, component: unknown) => () => void + +/** + * One rendered slot's local view, from {@link SlotTestRuntime.renderSlot}: + * the `data-slot` wrapper is the snapshot root (`expect(view.container) + * .toMatchSnapshot()` captures exactly this slot's output), Testing Library + * queries are bound inside it, and `update` re-renders with new owner props. + */ +export interface SlotView { + /** The `
` wrapper around the slot's rendered output. */ + readonly container: HTMLElement + /** Testing Library queries scoped to {@link SlotView.container}. */ + readonly view: BoundFunctions + /** + * Replace the owner props and flush the re-render (the render-site update: + * in production the owner recomputes the share and React re-renders). + * @param owner - the next owner props share. + */ + update(owner: OwnerOf): void +} + +/** + * Mounted feature plugin handle: the live fiber plus an act-wrapped, + * idempotent dispose (unload cascade: entries, declared child slots, store + * instances, and provided services all fall together). + */ +export interface FeatureHandle { + /** The plugin's live Cordis fiber (state assertions, escape hatch). */ + readonly fiber: Fiber + /** + * Dispose the plugin fiber inside React act; repeated calls no-op. + * @returns completion of the unload cascade. + */ + dispose(): Promise +} + +/** + * Owner-props cell behind the auto frame: one external store the frame + * subscribes to, so {@link SlotTestRuntime.renderSlot} and + * {@link SlotView.update} drive React through the standard uSES seam. + */ +class OwnerPropsCell { + private readonly owners = new Map() + private readonly listeners = new Set<() => void>() + private version = 0 + + /** Snapshot version for uSES pairing (bumped on every set). */ + readonly getVersion = (): number => this.version + + /** + * Subscribe to owner-props changes. + * @param fn - change callback. + * @returns unsubscribe. + */ + readonly subscribe = (fn: () => void): (() => void) => { + this.listeners.add(fn) + return () => { this.listeners.delete(fn) } + } + + /** + * Install or replace one key's owner props and notify (synchronous; the + * caller wraps in act). + * @param key - slot key. + * @param owner - owner props share. + */ + set(key: string, owner: object): void { + this.owners.set(key, owner) + this.version += 1 + for (const fn of [...this.listeners]) fn() + } + + /** Keys with supplied owner props, in first-supply order. */ + entries(): readonly (readonly [string, object])[] { + return [...this.owners.entries()] + } +} + +/** + * The test-owned 'root' occupant: declares the child slots a suite needs + * through the REAL `slots.register`, with a caller-supplied minimal frame — + * the runtime never guesses a feature's page structure. + */ +export class TestRoot { + private disposeEntry: (() => void) | undefined + + /** + * @param slots - the runtime SlotsService. + * @param stabilize - the owning runtime's act wrapper. + */ + constructor(private readonly slots: SlotsService, private readonly stabilize: Stabilizer) {} + + /** + * Register the root frame, declaring (and thereby claiming) the child + * slots. One declaration per runtime — a second call fails loud in the + * core ('root' is a single slot). + * @param children - child-slot declaration table (declaration + render authorization + runtime spec). + * @param frame - minimal frame component; its props derive from the declared keys (composed-props contract). + * @returns completion of the act-wrapped registration. + */ + async declare( + children: D, + frame: SlotComponent & keyof SlotMap & string, undefined, object>>, + ): Promise { + await this.stabilize(() => { + // Erased hop (same pattern as SlotsService's own implementation arm); + // the declare signature above is the typed seam. + this.disposeEntry = (this.slots.register as unknown as ErasedRegister)({ name: 'root', children }, frame) + }) + } + + /** Remove the root registration and collapse its declarations (runtime dispose path). */ + release(): void { + this.disposeEntry?.() + this.disposeEntry = undefined + } +} + +/** + * The assembled test runtime. Obtain via {@link SlotTestRuntime.create}; + * dispose with {@link SlotTestRuntime.dispose} (afterEach). Public mutators + * are act-wrapped throughout — tests never handle SlotCore microtask + * batching or React act themselves. + */ +export class SlotTestRuntime { + /** The runtime's Cordis root (escape hatch: extra services via `ctx.provide`, raw `ctx.plugin` mounts). */ + readonly ctx: Context + /** The production SlotsService mounted on {@link SlotTestRuntime.ctx}. */ + readonly slots: SlotsService + /** The test-owned 'root' occupant. */ + readonly root: TestRoot + /** Sessions double (list/current observable, cells, scopes, behavior faces). */ + readonly sessions: TestSessions + /** Workspaces double (list observable, recorded intent actions). */ + readonly workspaces: TestWorkspaces + + private readonly stabilizer: Stabilizer = async (fn) => { + await act(async () => { await fn() }) + } + + private host: SlotRendererHost | undefined + private readonly views: RenderResult[] = [] + private readonly handles: FeatureHandle[] = [] + private disposed = false + /** Auto-frame state ({@link SlotTestRuntime.declare} / {@link SlotTestRuntime.renderSlot}). */ + private readonly ownerCell = new OwnerPropsCell() + private readonly autoDeclared = new Set() + private autoRootView: RenderResult | undefined + + private constructor(ctx: Context, slots: SlotsService) { + this.ctx = ctx + this.slots = slots + this.root = new TestRoot(slots, this.stabilizer) + this.sessions = new TestSessions(this.stabilizer, ctx) + this.workspaces = new TestWorkspaces(this.stabilizer) + ctx.provide('sessions', this.sessions) + ctx.provide('workspaces', this.workspaces) + // Capturing install: the production renderer does the rendering; the + // wrapper only takes the host face for storeOf (no machinery copied). + const renderer = createSlotRenderer() + slots.install({ + renderRoot: (host, ownerProps) => { + this.host = host + return renderer.renderRoot(host, ownerProps) + }, + }) + } + + /** + * Assemble a runtime: real Context, mounted SlotsService, installed + * renderer, and the session/workspace doubles provided as services. + * @returns the ready runtime. + */ + static async create(): Promise { + registerDomSnapshotSerializer() + const ctx = new Context() + const fiber = ctx.plugin(SlotsService) + await fiber.await() + return new SlotTestRuntime(ctx, ctx.get('slots') as SlotsService) + } + + /** + * Provide an extra service the feature under test injects (e.g. a layout + * fake). Sugar over `ctx.provide`, typed against the Context declaration + * merge: for a declared service name the fake must be a subset of that + * service's outward face (Partial — supply only what the feature calls), + * so a production face change breaks the fake at compile time. Undeclared + * names stay unchecked (ad-hoc test services). + * @param name - service name. + * @param value - service implementation (test double). + */ + provide(name: K, value: K extends keyof Context ? Partial : unknown): void { + this.ctx.provide(name, value) + } + + /** + * Mount a feature plugin on a real fiber. Required services are prechecked + * so a missing provider fails loud instead of suspending the fiber forever + * (deliberate load-order suspension tests use `ctx.plugin` directly). + * @param plugin - plugin value (function, class, or `{ inject, apply }` object). + * @returns handle owning the fiber's explicit disposal. + */ + async mount(plugin: Plugin): Promise { + const required = Object.keys(Inject.resolve((plugin as { inject?: Inject }).inject)) + const missing = required.filter(name => this.ctx.get(name) === undefined) + if (missing.length > 0) { + throw new Error(`mount would suspend: missing service(s) ${missing.join(', ')} — provide() them first`) + } + const fiber = this.ctx.plugin(plugin) + await this.stabilizer(async () => { + await fiber.await() + }) + let disposed = false + const handle: FeatureHandle = { + fiber, + dispose: async () => { + if (disposed) return + disposed = true + await this.stabilizer(() => fiber.dispose()) + }, + } + this.handles.push(handle) + return handle + } + + /** + * Render the root slot tree through the ctx-level entry (the shell's own + * seam): `ctx.slots.renderSlot('root', {})` under Testing Library. + * @returns the Testing Library view. + */ + renderRoot(): RenderResult { + const view = render(createElement(Fragment, null, this.slots.renderSlot('root', {}))) + this.views.push(view) + return view + } + + /** + * Declare child slots under an auto-generated root frame — the single-slot + * mounting path for local DOM snapshots. Each key later supplied through + * {@link SlotTestRuntime.renderSlot} renders inside its own + * `
` wrapper (the snapshot root). Mutually exclusive + * with {@link TestRoot.declare} ('root' is a single slot); one call per + * runtime. + * @param children - child-slot declaration table (same contract as TestRoot.declare). + * @returns completion of the act-wrapped registration. + */ + async declare(children: ChildrenDecl): Promise { + for (const key of Object.keys(children)) this.autoDeclared.add(key) + const cell = this.ownerCell + const AutoFrame = (props: { renderSlot: (key: string, owner: object) => ReactNode }) => { + useSyncExternalStore(cell.subscribe, cell.getVersion) + return createElement(Fragment, null, cell.entries().map(([key, owner]) => + createElement('div', { 'data-slot': key, key }, props.renderSlot(key, owner)))) + } + await this.root.declare(children as never, AutoFrame as never) + } + + /** + * Render one declared slot with its owner props and return the local view. + * The whole root tree mounts through the production assembly path + * (renderer, scope providers, store axis); only this key's output lands in + * the returned container. Call again with another key to view a sibling + * slot of the same tree. + * @param key - a key declared through {@link SlotTestRuntime.declare}. + * @param owner - owner props share for the render site. + * @returns the slot-local view (snapshot container, scoped queries, owner updates). + */ + renderSlot(key: K, owner: OwnerOf): SlotView { + if (!this.autoDeclared.has(key)) { + throw new Error(`renderSlot('${key}') without declare() — declare the key first (or use root.declare for a custom frame)`) + } + const install = (next: object): void => { + // Synchronous cell write inside act: the frame re-renders through uSES. + act(() => { + this.ownerCell.set(key, next) + }) + } + install(owner) + this.autoRootView ??= this.renderRoot() + const container = this.autoRootView.container.querySelector(`[data-slot="${key}"]`) + if (!(container instanceof HTMLElement)) { + throw new Error(`renderSlot('${key}'): the auto frame rendered no wrapper — was the runtime already disposed?`) + } + return { container, view: within(container), update: install } + } + + /** + * Resolve the store instance the renderer would hand a slot's component + * (identity assertions, action-driven writes). Requires a prior + * {@link SlotTestRuntime.renderRoot} — the host face exists only inside the + * installed renderer, exactly as in production. + * @param key - slot key whose first entry declares the store. + * @param scopeKey - session id for session-scope slots; omit for root scope. + * @returns the live store instance. + */ + storeOf(key: keyof SlotMap & string, scopeKey?: string): StoreInstanceLike { + if (this.host === undefined) { + throw new Error('storeOf before renderRoot() — the host face exists only inside the installed renderer') + } + const entry = this.host.entriesOf(key)[0] + if (entry === undefined) throw new Error(`storeOf('${key}'): no registration on the ledger`) + const instance = this.host.storeOf(entry, scopeKey) + if (instance === undefined) throw new Error(`storeOf('${key}'): the entry declares no store`) + return instance + } + + /** + * Flush pending ledger/store notifications inside act — for mutations made + * outside the runtime's own methods (e.g. a direct `slots.register`). + * @returns completion of the act pass. + */ + async flush(): Promise { + await this.stabilizer(() => {}) + } + + /** + * Tear down: unmount React trees first, then dispose feature fibers, the + * root registration, minted session scopes, and persisted test state. + * Idempotent. + * @returns completion of the teardown. + */ + async dispose(): Promise { + if (this.disposed) return + this.disposed = true + this.autoRootView = undefined + for (const view of this.views.splice(0)) view.unmount() + for (const handle of this.handles.splice(0)) await handle.dispose() + this.root.release() + await this.sessions.disposeScopes() + localStorage.clear() + } +} diff --git a/packages/client/test-runtime/src/invariant.ts b/packages/client/test-runtime/src/invariant.ts new file mode 100644 index 0000000000..09ef3da6ce --- /dev/null +++ b/packages/client/test-runtime/src/invariant.ts @@ -0,0 +1,32 @@ +/** + * Package-owned invariant companion for `@deepseek-ai/dsh-client-test-runtime`. + * @module @deepseek-ai/dsh-client-test-runtime/invariant + */ + +/* jscpd:ignore-start */ +import type { Context } from 'cordis' +import type { InvariantInstaller } from '@deepseek-ai/dsh-invariants' + +const PACKAGE_NAME = '@deepseek-ai/dsh-client-test-runtime' + +/** Cordis companion plugin name. */ +export const name = 'client-test-runtime-invariant' +/** Service required before the companion can reserve package ownership. */ +export const inject = ['invariants'] + +/** + * No runtime invariant: this test-support package owns no production event + * stream or mutable data — it assembles the runtime SlotsService and renderer + * (whose packages own their invariants) around test doubles; its own behavior + * is exercised by its package tests. + */ +const install: InvariantInstaller = () => {} + +/** + * Register this package's invariant companion. + * @param ctx - Cordis context carrying the invariant service. + * @returns the installed registration's disposer after setup succeeds. + */ +export const apply = (ctx: Context): Promise<() => void> => + Promise.resolve(ctx.invariants.register(PACKAGE_NAME, install)) +/* jscpd:ignore-end */ diff --git a/packages/client/test-runtime/src/sessions.ts b/packages/client/test-runtime/src/sessions.ts new file mode 100644 index 0000000000..7bf7992105 --- /dev/null +++ b/packages/client/test-runtime/src/sessions.ts @@ -0,0 +1,398 @@ +/** Test-owned sessions face: the SlotsService host contract over declarative fixtures. */ +import type { Context } from 'cordis' +import { createScope, scopeOf, SessionProvideChannel } from '@deepseek-ai/dsh-client-runtime/client' +import { createSnapshotStore } from '@deepseek-ai/dsh-client-runtime/client' +import type { + ConversationSnapshot, ISessions, ObservableSnapshot, ProjectionsFace, SessionFace, SessionId, + SessionListState, SessionProvideDescriptor, SessionSummary, SnapshotStore, +} from '@deepseek-ai/dsh-client-runtime/client' +import type { HostObservable, SessionMaybeProvideInfo, SessionProvideInfo } from '@deepseek-ai/dsh-client-ui-slots' +import { conversationSnapshot } from './fixtures.ts' +import type { SessionFixture, Stabilizer } from './fixtures.ts' + +/** + * The fixture-backed session face: conversation reads delegate to the + * fixture's snapshot store; ISession verbs are fail-loud stubs unless the + * fixture supplies them (the runtime never fakes behavior a test did not + * declare — an unstubbed call names itself instead of half-working). Extra + * fixture methods are grafted verbatim for feature-side casts. + */ +export class FixtureSession implements SessionFace { + /** + * The useProjection seat: identity-stable per-key faces over the fixture's + * projection values (set via {@link TestSessions.setProjection}). + */ + readonly projections: ProjectionsFace & { set(key: string, value: unknown): void } + + /** + * @param sessionId - host identity (branded view of the fixture id). + * @param store - conversation snapshot store (updateSnapshot writes it). + * @param overrides - fixture-declared behavior face, grafted over the stubs. + */ + constructor( + readonly sessionId: SessionId, + private readonly store: SnapshotStore, + overrides: Record, + ) { + const values = new Map() + const listeners = new Map void>>() + const faces = new Map>() + this.projections = { + faceOf: (key: string) => { + let face = faces.get(key) + if (face === undefined) { + face = { + getSnapshot: () => values.get(key), + subscribe: (fn: () => void) => { + const set = listeners.get(key) ?? new Set() + set.add(fn) + listeners.set(key, set) + return () => { set.delete(fn) } + }, + } + faces.set(key, face) + } + return face + }, + set: (key: string, value: unknown) => { + values.set(key, value) + for (const fn of [...(listeners.get(key) ?? [])]) fn() + }, + } + Object.assign(this, overrides) + } + + /** @returns the fixture conversation snapshot (useSession read side). */ + getSnapshot(): ConversationSnapshot { + return this.store.getSnapshot() + } + + /** + * Subscribe to fixture snapshot changes. + * @param fn - change callback. + * @returns unsubscribe. + */ + subscribe(fn: () => void): () => void { + return this.store.subscribe(fn) + } + + /** + * Fail-loud stub; supply `prompt` on the fixture's session face to exercise it. + * @returns never — always throws. + */ + prompt(): never { + throw new Error(`test session "${this.sessionId}": prompt is not stubbed — supply it on the fixture's session face`) + } + + /** + * Fail-loud stub; supply `cancel` on the fixture's session face to exercise it. + * @returns never — always throws. + */ + cancel(): never { + throw new Error(`test session "${this.sessionId}": cancel is not stubbed — supply it on the fixture's session face`) + } + + /** + * Fail-loud stub; supply `loadOlder` on the fixture's session face to exercise it. + * @returns never — always throws. + */ + loadOlder(): never { + throw new Error(`test session "${this.sessionId}": loadOlder is not stubbed — supply it on the fixture's session face`) + } + + /** + * Fail-loud stub; supply `loadAllHistory` on the fixture's session face to exercise it. + * @returns never — always throws. + */ + loadAllHistory(): never { + throw new Error(`test session "${this.sessionId}": loadAllHistory is not stubbed — supply it on the fixture's session face`) + } +} + +/** One live test session: fixture-derived stores plus its minted scope state. */ +interface SessionRecord { + summary: SessionSummary + snapshot: SnapshotStore + session: FixtureSession + scope: Context | undefined + scopeFiber: { dispose(): Promise } | undefined + /** Materialized standard-props bundle (identity-stable per session; invalidated on roster change). */ + provideInfo: SessionProvideInfo | undefined +} + +/** Test binding shape handed to provider resolvers and feature injects (a SessionBinding whose session is the fixture face). */ +export interface TestSessionBinding { + readonly sessionId: SessionId + readonly session: FixtureSession + readonly ctx: Context +} + +/** + * Sessions test double behind the renderer host and feature injects: owns the + * list/current observable, the standard-props provide channel (the runtime's + * `useSession` contribution included), scope minting through the production + * `createScope`, and the session behavior face supplied per fixture. + * + * Implements the same ISessions face features receive as `ctx.sessions`, so + * a production face change breaks this double at compile time; the extra + * members (add/updateSnapshot/setCurrent/remove/behavior/calls and the + * legacy provideInfo/maybeProvideInfo lookups) are bench-only surface. + */ +export class TestSessions implements ISessions { + /** The useSessions standard feed (list rows + current selection). */ + readonly list: SnapshotStore + /** + * Atomic current-session provide projection (production SessionsService + * mirror): selection changes and provider-roster changes publish through + * this one source — the member the SlotsService host face hands the + * renderer's SessionProvider. + */ + readonly currentProvideInfo: HostObservable + private readonly records = new Map() + /** The production provide channel (roster, materialization rules, current projection) — no test-side mirror. */ + private readonly channel: SessionProvideChannel + + /** Calls observed on the service-level face (open/clear), newest last. */ + readonly calls: { method: 'open' | 'clear'; args: unknown[] }[] = [] + + /** + * @param stabilize - the owning runtime's act wrapper. + * @param rootCtx - the runtime's Cordis root; scope fibers mount under it. + */ + constructor(private readonly stabilize: Stabilizer, private readonly rootCtx: Context) { + this.list = createSnapshotStore({ + ids: [], byId: {}, current: undefined, phase: 'ready', + }) + this.channel = new SessionProvideChannel({ + rebuildBundles: () => { + for (const record of this.records.values()) { + if (record.provideInfo !== undefined) { + record.provideInfo = this.channel.materializeInfo(this.bindingOf(record.session.sessionId, record)) + } + } + }, + resolveCurrent: () => this.maybeProvideInfo(this.list.getSnapshot().current), + }) + this.currentProvideInfo = this.channel.currentProvideInfo + // The projection follows every current write, as in production. + this.list.subscribe(() => { this.channel.publishCurrent() }) + } + + /** + * Add a session from a fixture and (by default) make it current. + * @param fixture - identity + snapshot/summary overrides + behavior face. + * @param opts - pass `current: false` to add without selecting. + * @returns the stable session id (branded view of `fixture.id`). + */ + async add(fixture: SessionFixture, opts?: { current?: boolean }): Promise { + const id = fixture.id as SessionId + if (this.records.has(id)) throw new Error(`test session "${id}" already added`) + const summary: SessionSummary = { + id, + displayTitle: fixture.id, + running: false, + blank: false, + updatedAt: this.records.size + 1, + ...fixture.summary, + } + const snapshot = createSnapshotStore({ + ...conversationSnapshot(id), + ...fixture.snapshot, + }) + this.records.set(id, { + summary, + snapshot, + session: new FixtureSession(id, snapshot, fixture.session ?? {}), + scope: undefined, + scopeFiber: undefined, + provideInfo: undefined, + }) + await this.stabilize(() => { + this.list.update((draft) => { + draft.ids.push(id) + draft.byId[id] = summary + if (opts?.current !== false) draft.current = id + }) + }) + return id + } + + /** + * Update a session's conversation snapshot through an immer draft (the + * live-stream stand-in: components subscribed via useSession re-render). + * @param id - session id. + * @param mutate - draft mutator. + */ + async updateSnapshot(id: string, mutate: (draft: ConversationSnapshot) => void): Promise { + const record = this.require(id) + await this.stabilize(() => { record.snapshot.update(mutate) }) + } + + /** + * Switch the current selection (undefined = the no-session empty state). + * @param id - session id to select, or undefined to clear. + */ + async setCurrent(id: string | undefined): Promise { + if (id !== undefined) this.require(id) + await this.stabilize(() => { + this.list.update((draft) => { draft.current = id as SessionId | undefined }) + }) + } + + /** + * Remove a session: list row, scope fiber, and per-session store instances + * (with persisted state) die together — the same single lifecycle axis the + * production SessionsService drives on session death, minus staging. + * @param id - session id. + */ + async remove(id: string): Promise { + const record = this.require(id) + this.records.delete(id as SessionId) + await this.stabilize(async () => { + this.list.update((draft) => { + draft.ids = draft.ids.filter(existing => existing !== id) + const { [id as SessionId]: _dead, ...rest } = draft.byId + draft.byId = rest + if (draft.current === id) draft.current = undefined + }) + if (record.scopeFiber !== undefined) await record.scopeFiber.dispose() + this.rootCtx.get('slots')?.pruneStoreScope(id) + }) + } + + /** + * Register a per-session standard-props provider (production `provide` + * contract: hooks become `use` selector hooks on the render side, + * props spread verbatim; duplicate names fail loud at materialization). + * @param descriptor - static member roster plus per-session resolver. + * @returns disposer removing the provider. + */ + provide(descriptor: SessionProvideDescriptor): () => void { + return this.channel.provide(descriptor) + } + + /** + * Resolve the definite per-session standard-props bundle (host face member). + * @param id - session id. + * @returns the identity-stable bundle, or undefined for unknown sessions. + */ + provideInfo(id: string): SessionProvideInfo | undefined { + const record = this.records.get(id as SessionId) + if (record === undefined) return undefined + record.provideInfo ??= this.channel.materializeInfo(this.bindingOf(id as SessionId, record)) + return record.provideInfo + } + + /** + * Resolve the current-session-optional standard kit (host face member): + * unknown or absent ids return the static no-session projection. + * @param id - current session id, when selected. + * @returns a definite or no-session provide bundle. + */ + maybeProvideInfo(id: string | undefined): SessionMaybeProvideInfo { + return (id === undefined ? undefined : this.provideInfo(id)) ?? this.channel.maybeInfo + } + + /** + * Resolve (mint on first touch) the session-scoped Cordis context through + * the production `createScope`, so real `scopeOf`/scope-addressed services + * resolve it. + * @param id - session id. + * @returns the scoped context, or undefined for unknown sessions. + */ + scope(id: string): Context | undefined { + const record = this.records.get(id as SessionId) + if (record === undefined) return undefined + if (record.scope === undefined) { + const handle = createScope(this.rootCtx, id as SessionId) + record.scope = handle.ctx + record.scopeFiber = handle.fiber + } + return record.scope + } + + /** + * Session assembly binding (inject factories and provide resolvers receive it). + * @param id - session id. + * @returns sessionId + behavior face + scoped ctx, or undefined when unknown. + */ + binding(id: string): TestSessionBinding | undefined { + const record = this.records.get(id as SessionId) + if (record === undefined) return undefined + return this.bindingOf(id as SessionId, record) + } + + /** + * Read the session scope tag off a context (service-method seam mirror). + * @param ctx - any client context. + * @returns the session id, or undefined on root contexts. + */ + scopeOf(ctx: Context): SessionId | undefined { + return scopeOf(ctx) + } + + /** + * Resolve the scoped session face off a context (production `sessionOf` + * mirror). + * @param ctx - any client context. + * @returns the fixture session face, or undefined off-scope. + */ + sessionOf(ctx: Context): SessionFace | undefined { + const id = scopeOf(ctx) + if (id === undefined) return undefined + return this.records.get(id)?.session + } + + /** + * Service-level selection call (recorded, then applied to the list store + * synchronously — inject callbacks call this outside any act window; the + * store notify is microtask-batched so the next stabilized step observes it). + * @param id - session id. + */ + open(id: SessionId): void { + this.calls.push({ method: 'open', args: [id] }) + this.require(id) + this.list.update((draft) => { draft.current = id }) + } + + /** Clear the current selection (recorded; the production no-session flow). */ + clear(): void { + this.calls.push({ method: 'clear', args: [] }) + this.list.update((draft) => { draft.current = undefined }) + } + + /** + * The session face of a fixture (typed view for assertions; fixture + * behavior methods are grafted onto it). + * @param id - session id. + * @returns the FixtureSession the binding and provide channel carry. + */ + behavior(id: string): FixtureSession { + return this.require(id).session + } + + /** Dispose minted scope fibers (runtime dispose path). */ + async disposeScopes(): Promise { + for (const record of this.records.values()) { + if (record.scopeFiber !== undefined) { + await record.scopeFiber.dispose() + record.scope = undefined + record.scopeFiber = undefined + } + } + } + + private bindingOf(id: SessionId, record: SessionRecord): TestSessionBinding { + const ctx = this.scope(id) + /* v8 ignore next 2 -- bindingOf only runs for a live record, whose scope + * always resolves; kept so a future caller cannot mint a ctx-less binding. */ + if (ctx === undefined) throw new Error(`test session "${id}" resolved no scope`) + return { sessionId: id, session: record.session, ctx } + } + + private require(id: string): SessionRecord { + const record = this.records.get(id as SessionId) + if (record === undefined) throw new Error(`test session "${id}" is not added`) + return record + } +} diff --git a/packages/client/test-runtime/src/snapshot.ts b/packages/client/test-runtime/src/snapshot.ts new file mode 100644 index 0000000000..0165676cdf --- /dev/null +++ b/packages/client/test-runtime/src/snapshot.ts @@ -0,0 +1,89 @@ +/** + * DOM snapshot hygiene: a vitest snapshot serializer that keeps `.snap` + * files structural. Two normalizations, both on a clone (the live DOM is + * untouched, so class/tag queries keep working): + * + * - CSS-module scoped class names (`_frame_334d2d`, this repo's + * `_[local]_[hash]` shape) fold back to their semantic local (`frame`), so + * CSS edits do not churn snapshots. + * - `` internals collapse to a `data-content` fingerprint on the svg + * element: path geometry is print noise, but the fingerprint still flips + * when an icon's artwork actually changes. + */ +import { expect } from 'vitest' +import type { SnapshotSerializer } from 'vitest' + +/** One scoped class token: `__` (local may itself contain underscores). */ +const SCOPED_CLASS = /^_(.+)_[a-z0-9]+$/ + +/** Fold scoped tokens in one class attribute value; foreign tokens pass through. */ +function normalizeClassValue(value: string): string { + return value + .split(/\s+/) + .filter(token => token !== '') + .map(token => token.replace(SCOPED_CLASS, '$1')) + .join(' ') +} + +/** FNV-1a 32-bit over the svg markup: deterministic, dependency-free fingerprint. */ +function fingerprint(markup: string): string { + let hash = 0x811c9dc5 + for (let i = 0; i < markup.length; i++) { + hash ^= markup.charCodeAt(i) + hash = Math.imul(hash, 0x01000193) + } + return (hash >>> 0).toString(16).padStart(8, '0') +} + +/** svg elements of a subtree, the root included when it is one. */ +function svgsOf(root: Element): Element[] { + const svgs: Element[] = [...root.querySelectorAll('svg')] + if (root.tagName.toLowerCase() === 'svg') svgs.unshift(root) + return svgs +} + +/** Whether serializing this subtree needs a normalized clone. */ +function needsNormalization(root: Element): boolean { + const scoped = [root, ...root.querySelectorAll('[class]')].some((el) => { + const value = el.getAttribute('class') + return value !== null && value.split(/\s+/).some(token => SCOPED_CLASS.test(token)) + }) + return scoped || svgsOf(root).some(svg => svg.childNodes.length > 0) +} + +/** + * The serializer plugin. Matches DOM elements whose subtree carries a scoped + * class or svg internals; serializes a normalized clone, which no longer + * matches, so printing falls through to the built-in DOM element serializer. + */ +export const domSnapshotSerializer: SnapshotSerializer = { + test(value: unknown): boolean { + return typeof Element !== 'undefined' && value instanceof Element && needsNormalization(value) + }, + serialize(value, config, indentation, depth, refs, printer): string { + const clone = (value as Element).cloneNode(true) as Element + for (const el of [clone, ...clone.querySelectorAll('[class]')]) { + const raw = el.getAttribute('class') + if (raw !== null) el.setAttribute('class', normalizeClassValue(raw)) + } + for (const svg of svgsOf(clone)) { + if (svg.childNodes.length === 0) continue + svg.setAttribute('data-content', fingerprint(svg.innerHTML)) + svg.replaceChildren() + } + return printer(clone, config, indentation, depth, refs) + }, +} + +let registered = false + +/** + * Register {@link domSnapshotSerializer} with vitest's expect (idempotent). + * SlotTestRuntime.create() calls this; specs that snapshot DOM outside the + * runtime import and call it themselves. + */ +export function registerDomSnapshotSerializer(): void { + if (registered) return + registered = true + expect.addSnapshotSerializer(domSnapshotSerializer) +} diff --git a/packages/client/test-runtime/src/workspaces.ts b/packages/client/test-runtime/src/workspaces.ts new file mode 100644 index 0000000000..32c9bff36e --- /dev/null +++ b/packages/client/test-runtime/src/workspaces.ts @@ -0,0 +1,147 @@ +/** Test-owned workspaces face: the renderer standard-kit observable plus recorded actions. */ +import { createSnapshotStore } from '@deepseek-ai/dsh-client-runtime/client' +import type { + IWorkspaces, SessionId, SnapshotStore, WorkspaceId, WorkspaceListState, WorkspaceView, +} from '@deepseek-ai/dsh-client-runtime/client' +import { workspaceListState } from './fixtures.ts' +import type { Stabilizer } from './fixtures.ts' + +/** + * Workspaces test double. Implements the same IWorkspaces face features + * receive as `ctx.workspaces`, so a production face change breaks this + * double at compile time. Every action records into {@link + * TestWorkspaces.calls}; defaults are inert echoes — feature tests needing + * richer behavior replace them via {@link TestWorkspaces.stub}. + */ +export class TestWorkspaces implements IWorkspaces { + /** The useWorkspaces standard feed. */ + readonly list: SnapshotStore + + /** Calls observed on the action face, newest last. */ + readonly calls: { method: string; args: unknown[] }[] = [] + + /** Replaceable action seat: feature tests may stub richer behavior. */ + private readonly stubs = new Map unknown>() + + /** + * @param stabilize - the owning runtime's act wrapper. + */ + constructor(private readonly stabilize: Stabilizer) { + this.list = createSnapshotStore(workspaceListState()) + } + + /** + * Update the workspace list state through an immer draft. + * @param mutate - draft mutator. + */ + async update(mutate: (draft: WorkspaceListState) => void): Promise { + await this.stabilize(() => { this.list.update(mutate) }) + } + + /** + * Replace an action's behavior (the recorded call is still appended first). + * @param method - action name (e.g. 'connectWorkspace'). + * @param impl - replacement behavior. + */ + stub(method: string, impl: (...args: unknown[]) => unknown): void { + this.stubs.set(method, impl) + } + + /** + * Connect a workspace to its reusable/new blank session (recorded). The + * default resolves the workspace id back as the session id; stub for + * cross-session flows. + * @param workspaceId - target workspace. + * @returns the connected session id. + */ + async connectWorkspace(workspaceId: WorkspaceId): Promise { + this.calls.push({ method: 'connectWorkspace', args: [workspaceId] }) + const stub = this.stubs.get('connectWorkspace') + if (stub !== undefined) return await (stub(workspaceId) as Promise) + return `session-of-${workspaceId}` as SessionId + } + + /** + * New-session flow (recorded; stubbed behavior runs when installed). + * @param workspaceId - optional explicit workspace target. + */ + startSession(workspaceId?: WorkspaceId): void { + this.calls.push({ method: 'startSession', args: [workspaceId] }) + this.stubs.get('startSession')?.(workspaceId) + } + + /** + * Create a Workspace (recorded). The default echoes a view derived from + * the input; stub for failure or list-coupled flows. + * @param input - exactly one Host create spelling. + * @returns the created Workspace view. + */ + async create(input: { name: string } | { path: string }): Promise { + this.calls.push({ method: 'create', args: [input] }) + const stub = this.stubs.get('create') + if (stub !== undefined) return await (stub(input) as Promise) + const title = 'name' in input ? input.name : input.path + return { + workspaceId: `ws-${title}` as WorkspaceId, + title, + path: 'path' in input ? input.path : `/${input.name}`, + sessionIds: [], + } as unknown as WorkspaceView + } + + /** + * Open a path with the host OS default application (recorded; default no-op). + * @param path - host-resolvable path. + */ + async openPath(path: string): Promise { + this.calls.push({ method: 'openPath', args: [path] }) + await (this.stubs.get('openPath')?.(path) as Promise | undefined) + } + + /** + * Directory picker (recorded). The default cancels (null); stub to select. + * @returns the picked path, or null. + */ + async pickDirectory(): Promise { + this.calls.push({ method: 'pickDirectory', args: [] }) + const stub = this.stubs.get('pickDirectory') + if (stub !== undefined) return await (stub() as Promise) + return null + } + + /** + * Rename a Workspace (recorded). The default echoes a minimal view. + * @param workspaceId - target workspace. + * @param title - new title. + * @returns the updated view. + */ + async rename(workspaceId: WorkspaceId, title: string): Promise { + this.calls.push({ method: 'rename', args: [workspaceId, title] }) + const stub = this.stubs.get('rename') + if (stub !== undefined) return await (stub(workspaceId, title) as Promise) + return { workspaceId, title, path: `/${title}`, sessionIds: [] } as unknown as WorkspaceView + } + + /** + * Delete a Workspace (recorded; default no-op). + * @param workspaceId - target workspace. + */ + async delete(workspaceId: WorkspaceId): Promise { + this.calls.push({ method: 'delete', args: [workspaceId] }) + await (this.stubs.get('delete')?.(workspaceId) as Promise | undefined) + } + + /** + * Move an accounted session (recorded). The default echoes a minimal view. + * @param workspaceId - target workspace. + * @param sessionId - session to move. + * @param beforeSessionId - anchor; omitted appends. + * @returns the updated view. + */ + async insertSessionBefore(workspaceId: WorkspaceId, sessionId: SessionId, beforeSessionId?: SessionId): Promise { + this.calls.push({ method: 'insertSessionBefore', args: [workspaceId, sessionId, beforeSessionId] }) + const stub = this.stubs.get('insertSessionBefore') + if (stub !== undefined) return await (stub(workspaceId, sessionId, beforeSessionId) as Promise) + return { workspaceId, title: '', path: '', sessionIds: [sessionId] } as unknown as WorkspaceView + } +} diff --git a/packages/client/test-runtime/tests/__snapshots__/runtime.spec.tsx.snap b/packages/client/test-runtime/tests/__snapshots__/runtime.spec.tsx.snap new file mode 100644 index 0000000000..3b46019a43 --- /dev/null +++ b/packages/client/test-runtime/tests/__snapshots__/runtime.spec.tsx.snap @@ -0,0 +1,37 @@ +// Vitest Snapshot v1, https://vitest.dev/guide/snapshot.html + +exports[`single-slot mounting (declare + renderSlot) > folds class hashes and collapses svg internals in snapshots, leaving the live DOM alone 1`] = ` +
+
+ + styled + + +
+
+`; + +exports[`single-slot mounting edge arms > serializes childless svg untouched next to scoped classes 1`] = ` +
+
+ +
+
+`; diff --git a/packages/client/test-runtime/tests/invariant.spec.ts b/packages/client/test-runtime/tests/invariant.spec.ts new file mode 100644 index 0000000000..837559ec81 --- /dev/null +++ b/packages/client/test-runtime/tests/invariant.spec.ts @@ -0,0 +1,12 @@ +import { describe, expect, it } from 'vitest' +import { Context } from 'cordis' +import * as TestRuntimeInvariant from '@deepseek-ai/dsh-client-test-runtime/invariant' +import InvariantService from '@deepseek-ai/dsh-invariants' + +describe('invariant companion', () => { + it('registers under the package name with an empty installer', async () => { + const ctx = new Context() + await ctx.plugin(InvariantService, { enabled: true }) + await expect(ctx.plugin(TestRuntimeInvariant).await()).resolves.toBeDefined() + }) +}) diff --git a/packages/client/test-runtime/tests/runtime.spec.tsx b/packages/client/test-runtime/tests/runtime.spec.tsx new file mode 100644 index 0000000000..1c53aa8033 --- /dev/null +++ b/packages/client/test-runtime/tests/runtime.spec.tsx @@ -0,0 +1,547 @@ +// @vitest-environment jsdom +/** + * SlotTestRuntime behavior: root declaration + rendering, session + * add/update/switch/remove through the real renderer, shared store identity + * and scope pruning, feature mount/dispose cascade, and runtime disposal + * idempotence. All through the production SlotsService + createSlotRenderer + * stack — this suite is the fixture the migrated feature specs rely on. + */ +import { afterEach, describe, expect, it, vi } from 'vitest' +import { cleanup } from '@testing-library/react' +import { defineStore } from '@deepseek-ai/dsh-client-runtime/client' +import type { SessionId, WorkspaceId } from '@deepseek-ai/dsh-client-runtime/client' +import type { PropsRenderSlots, SessionStandardProps } from '@deepseek-ai/dsh-client-ui-slots' +import { SlotTestRuntime } from '@deepseek-ai/dsh-client-test-runtime' + +declare module '@deepseek-ai/dsh-client-ui-slots' { + interface SlotMap { + 'trt.panel': { kind: 'single'; scope: 'root'; owner: { label?: string } } + 'trt.chat': { kind: 'single'; scope: 'session' } + 'trt.rows': { kind: 'list'; scope: 'root' } + } +} + +afterEach(cleanup) + +type FrameProps = PropsRenderSlots<'trt.panel' | 'trt.chat' | 'trt.rows'> + +/** Root frame declaring all three suite slots (render sites for each kind). */ +function Frame({ renderSlot, SessionProvider }: FrameProps) { + return ( + <> + {renderSlot('trt.panel', { label: 'from-owner' }, { fallback: no panel })} + no session}> + {() => renderSlot('trt.chat', {})} + + {renderSlot('trt.rows', {})} + + ) +} + +const CHILDREN = { + 'trt.panel': { kind: 'single', scope: 'root' }, + 'trt.chat': { kind: 'single', scope: 'session' }, + 'trt.rows': { kind: 'list', scope: 'root' }, +} as const + +async function runtimeWithFrame() { + const runtime = await SlotTestRuntime.create() + await runtime.root.declare(CHILDREN, Frame) + return runtime +} + +describe('root declaration and rendering', () => { + it('renders declared slots through the real renderer: fallback, then a live registration, then unload', async () => { + const runtime = await runtimeWithFrame() + const view = runtime.renderRoot() + expect(view.container.textContent).toContain('no panel') + + let dispose = (): void => {} + await runtime.flush() // no-op guard: flush outside mutations is safe + await (async () => { + dispose = runtime.slots.register( + { name: 'trt.panel' }, + ({ label }: { label?: string }) => panel:{label}) + await runtime.flush() + })() + expect(view.container.textContent).toContain('panel:from-owner') + dispose() + await runtime.flush() + expect(view.container.textContent).toContain('no panel') + await runtime.dispose() + }) + + it('fails loud when rendering with no root declaration (production boot-order check)', async () => { + const runtime = await SlotTestRuntime.create() + expect(() => runtime.renderRoot()).toThrow(/'root' has no registration/) + await runtime.dispose() + }) +}) + +describe('sessions', () => { + it('drives SessionProvider: empty state, current session, switch, live snapshot updates', async () => { + const runtime = await runtimeWithFrame() + runtime.slots.register({ name: 'trt.chat' }, (props: SessionStandardProps) => { + const running = props.useSession(s => s.running) + return chat:{props.sessionId}:{String(running)} + }) + const view = runtime.renderRoot() + expect(view.container.textContent).toContain('no session') + + await runtime.sessions.add({ id: 's1' }) + expect(view.container.textContent).toContain('chat:s1:false') + + await runtime.sessions.updateSnapshot('s1', (draft) => { draft.running = true }) + expect(view.container.textContent).toContain('chat:s1:true') + + await runtime.sessions.add({ id: 's2' }) // becomes current by default + expect(view.container.textContent).toContain('chat:s2:false') + + await runtime.sessions.setCurrent(undefined) + expect(view.container.textContent).toContain('no session') + await runtime.sessions.setCurrent('s1') + expect(view.container.textContent).toContain('chat:s1:true') + await runtime.dispose() + }) + + it('add with current:false keeps the selection; unknown ids fail loud on the mutators', async () => { + const runtime = await runtimeWithFrame() + await runtime.sessions.add({ id: 's1' }) + await runtime.sessions.add({ id: 's2' }, { current: false }) + expect(runtime.sessions.list.getSnapshot().current).toBe('s1') + expect(runtime.sessions.list.getSnapshot().ids).toEqual(['s1', 's2']) + await expect(runtime.sessions.add({ id: 's1' })).rejects.toThrow(/already added/) + await expect(runtime.sessions.setCurrent('ghost')).rejects.toThrow(/not added/) + await expect(runtime.sessions.updateSnapshot('ghost', () => {})).rejects.toThrow(/not added/) + await expect(runtime.sessions.remove('ghost')).rejects.toThrow(/not added/) + expect(() => runtime.sessions.behavior('ghost')).toThrow(/not added/) + await runtime.dispose() + }) + + it('mints REAL-tag scopes lazily and resolves them through the production scopeOf; bindings expose the behavior face', async () => { + const runtime = await runtimeWithFrame() + const prompt = vi.fn() + await runtime.sessions.add({ id: 's1', session: { prompt } }) + + expect(runtime.sessions.provideInfo('ghost')).toBeUndefined() + expect(runtime.sessions.scope('ghost')).toBeUndefined() + expect(runtime.sessions.binding('ghost')).toBeUndefined() + + const scope = runtime.sessions.scope('s1')! + expect(runtime.sessions.scope('s1')).toBe(scope) // stable per session + expect(runtime.sessions.scopeOf(scope)).toBe('s1') + expect(runtime.sessions.scopeOf(runtime.ctx)).toBeUndefined() + // sessionOf resolves the behavior face off the scope tag. + expect(runtime.sessions.sessionOf(scope)).toBe(runtime.sessions.behavior('s1')) + expect(runtime.sessions.sessionOf(runtime.ctx)).toBeUndefined() + + const binding = runtime.sessions.binding('s1')! + expect(binding.sessionId).toBe('s1') + expect(binding.ctx).toBe(scope) + ;(binding.session as { prompt: () => void }).prompt() + expect(prompt).toHaveBeenCalledOnce() + expect(runtime.sessions.behavior('s1')).toBe(binding.session) + // The binding's session doubles as the conversation observable face. + expect((binding.session as { getSnapshot(): { sessionId: string } }).getSnapshot().sessionId).toBe('s1') + + // A scoped service resolves through the scope ctx (scope-addressed pattern). + runtime.provide('probe', { hello: 'world' }) + expect(scope.get('probe')).toEqual({ hello: 'world' }) + await runtime.dispose() + }) + + it('materializes provide bundles: built-in session hook, custom providers, no-session projection', async () => { + const runtime = await runtimeWithFrame() + await runtime.sessions.add({ id: 's1' }) + + const info = runtime.sessions.provideInfo('s1')! + expect(info.sessionId).toBe('s1') + expect(info.hooks['session']).toBeDefined() // the built-in useSession source + expect(runtime.sessions.provideInfo('s1')).toBe(info) // identity-stable + + // A feature provider (the ui-conversation input pattern): declared names + // materialize per session and land in the no-session roster as undefined. + const off = runtime.sessions.provide({ + hooks: ['probe'], + props: ['probeActions'], + resolve: binding => ({ + hooks: { probe: { getSnapshot: () => binding.sessionId, subscribe: () => () => {} } }, + props: { probeActions: { poke: () => {} } }, + }), + }) + const enriched = runtime.sessions.provideInfo('s1')! + expect(enriched.hooks['probe']?.getSnapshot()).toBe('s1') + expect(enriched.props['probeActions']).toBeDefined() + const maybe = runtime.sessions.maybeProvideInfo(undefined) + expect(maybe.sessionId).toBeUndefined() + expect(Object.keys(maybe.hooks)).toEqual(['session', 'probe']) + expect(runtime.sessions.maybeProvideInfo('s1')).toBe(runtime.sessions.provideInfo('s1')) + expect(runtime.sessions.maybeProvideInfo('ghost').sessionId).toBeUndefined() + + // Misdeclared providers fail loud AT REGISTRATION (the production + // channel rebuilds live bundles eagerly and rolls the roster back): + // missing hook, missing prop, duplicate hook, duplicate prop. + expect(() => runtime.sessions.provide({ hooks: ['void'], resolve: () => ({}) })) + .toThrow(/missing hook "void"/) + expect(() => runtime.sessions.provide({ props: ['void'], resolve: () => ({}) })) + .toThrow(/missing prop "void"/) + expect(() => runtime.sessions.provide({ + hooks: ['session'], + resolve: () => ({ hooks: { session: { getSnapshot: () => 0, subscribe: () => () => {} } } }), + })).toThrow(/duplicate hook "session"/) + const propA = runtime.sessions.provide({ props: ['twice'], resolve: () => ({ props: { twice: 1 } }) }) + expect(() => runtime.sessions.provide({ props: ['twice'], resolve: () => ({ props: { twice: 2 } }) })) + .toThrow(/duplicate prop "twice"/) + propA() + // The rejected registrations rolled back: the roster still materializes. + expect(runtime.sessions.provideInfo('s1')).toBeDefined() + off() + off() // disposer is idempotent + expect(Object.keys(runtime.sessions.maybeProvideInfo(undefined).hooks)).toEqual(['session']) + await runtime.dispose() + }) + + it('records service-face calls; open() moves the selection and clear() empties it', async () => { + const runtime = await runtimeWithFrame() + await runtime.sessions.add({ id: 's1' }) + await runtime.sessions.add({ id: 's2' }) + runtime.sessions.open('s1' as SessionId) + await runtime.flush() + expect(runtime.sessions.list.getSnapshot().current).toBe('s1') + runtime.sessions.clear() + await runtime.flush() + expect(runtime.sessions.list.getSnapshot().current).toBeUndefined() + expect(runtime.sessions.calls).toEqual([ + { method: 'open', args: ['s1'] }, + { method: 'clear', args: [] }, + ]) + await runtime.dispose() + }) +}) + +describe('stores', () => { + const createSuiteStore = () => defineStore({ + init: () => ({ note: '' }), + persist: 'trt.store', + actions: { setNote: (d, note: string) => { d.note = note } }, + }) + + it('resolves per-session instances via the host face: shared identity, isolation, action-driven re-render', async () => { + const runtime = await runtimeWithFrame() + const handle = createSuiteStore() + runtime.slots.register( + { name: 'trt.chat', store: handle }, + (props: SessionStandardProps & { useStore: (sel: (s: { note: string }) => S) => S }) => + note:{props.useStore(s => s.note)}) + const view = runtime.renderRoot() + await runtime.sessions.add({ id: 's1' }) + + expect(() => runtime.storeOf('trt.panel')).toThrow(/no registration/) + const store = runtime.storeOf('trt.chat', 's1') + await runtime.flush() + ;(store.actions['setNote'] as (note: string) => void)('hello') + await runtime.flush() + expect(view.container.textContent).toContain('note:hello') + expect(runtime.storeOf('trt.chat', 's1')).toBe(store) // cached per scope key + + await runtime.sessions.add({ id: 's2' }) + const other = runtime.storeOf('trt.chat', 's2') + expect(other).not.toBe(store) + expect(other.getSnapshot()).toEqual({ note: '' }) + await runtime.dispose() + }) + + it('storeOf guards: before renderRoot, and for storeless entries', async () => { + const runtime = await runtimeWithFrame() + runtime.slots.register({ name: 'trt.panel' }, () => null) + expect(() => runtime.storeOf('trt.panel')).toThrow(/before renderRoot/) + runtime.renderRoot() + expect(() => runtime.storeOf('trt.panel')).toThrow(/declares no store/) + await runtime.dispose() + }) + + it('remove() prunes the session store scope: persisted state clears, a re-added session starts fresh', async () => { + const runtime = await runtimeWithFrame() + const handle = createSuiteStore() + runtime.slots.register({ name: 'trt.chat', store: handle }, () => null) + runtime.renderRoot() + await runtime.sessions.add({ id: 's1' }) + + const doomed = runtime.storeOf('trt.chat', 's1') + ;(doomed.actions['setNote'] as (note: string) => void)('buried') + expect(localStorage.getItem('trt.store.s1')).not.toBeNull() + + await runtime.sessions.remove('s1') + expect(localStorage.getItem('trt.store.s1')).toBeNull() + expect(runtime.sessions.list.getSnapshot().ids).toEqual([]) + expect(runtime.sessions.provideInfo('s1')).toBeUndefined() + + await runtime.sessions.add({ id: 's1' }) + const reborn = runtime.storeOf('trt.chat', 's1') + expect(reborn).not.toBe(doomed) + expect(reborn.getSnapshot()).toEqual({ note: '' }) + await runtime.dispose() + }) + + it('remove() also disposes a minted scope fiber; removing a non-current session keeps the selection', async () => { + const runtime = await runtimeWithFrame() + await runtime.sessions.add({ id: 's1' }) + await runtime.sessions.add({ id: 's2' }, { current: false }) + const scope = runtime.sessions.scope('s1')! + await runtime.sessions.remove('s2') + expect(runtime.sessions.list.getSnapshot().current).toBe('s1') + await runtime.sessions.remove('s1') + expect(scope.fiber.uid).toBeNull() // disposed fiber loses its uid + expect(runtime.sessions.list.getSnapshot().current).toBeUndefined() + await runtime.dispose() + }) +}) + +describe('workspaces', () => { + it('feeds useWorkspaces and records/stubs intent actions', async () => { + const runtime = await runtimeWithFrame() + runtime.slots.register( + { name: 'trt.panel' }, + (props: { useWorkspaces: (sel: (s: { phase: string }) => S) => S }) => + ws:{props.useWorkspaces(s => s.phase)}) + const view = runtime.renderRoot() + expect(view.container.textContent).toContain('ws:ready') + + await runtime.workspaces.update((draft) => { draft.phase = 'pending' }) + expect(view.container.textContent).toContain('ws:pending') + + runtime.workspaces.startSession('w1' as WorkspaceId) + await expect(runtime.workspaces.connectWorkspace('w2' as WorkspaceId)).resolves.toBe('session-of-w2') + expect(runtime.workspaces.calls).toEqual([ + { method: 'startSession', args: ['w1'] }, + { method: 'connectWorkspace', args: ['w2'] }, + ]) + const stub = vi.fn(() => Promise.resolve('other' as never)) + runtime.workspaces.stub('connectWorkspace', stub) + await expect(runtime.workspaces.connectWorkspace('w3' as WorkspaceId)).resolves.toBe('other') + expect(stub).toHaveBeenCalledOnce() + await runtime.dispose() + }) +}) + +describe('feature mount and disposal', () => { + it('mounts a plugin on a real fiber; dispose() cascades entries, declared children, and services', async () => { + const runtime = await runtimeWithFrame() + runtime.provide('layout', { openDetails: vi.fn() }) + const feature = await runtime.mount({ + inject: ['slots', 'layout'], + apply: (ctx: typeof runtime.ctx) => { + ctx.provide('feature-service', { ok: true }) + ctx.slots.register({ + name: 'trt.rows', + id: 'row-1', + children: { 'trt.rows.hole': { kind: 'single', scope: 'root' } }, + } as never, ((props: { renderSlot: (key: string, owner: object) => unknown }) => +
{props.renderSlot('trt.rows.hole', {}) as React.ReactNode}
) as never) + }, + }) + const view = runtime.renderRoot() + expect(view.getByTestId('row')).toBeTruthy() + expect(runtime.ctx.get('feature-service')).toEqual({ ok: true }) + expect(runtime.slots.entries('trt.rows')).toHaveLength(1) + + await feature.dispose() + await feature.dispose() // idempotent + expect(runtime.slots.entries('trt.rows')).toHaveLength(0) + expect(runtime.slots.spec('trt.rows.hole' as never)).toBeUndefined() + expect(runtime.ctx.get('feature-service')).toBeUndefined() + expect(view.queryByTestId('row')).toBeNull() + await runtime.dispose() + }) + + it('mount fails loud on missing services instead of suspending forever', async () => { + const runtime = await runtimeWithFrame() + await expect(runtime.mount({ inject: ['slots', 'absent-service'], apply: () => {} })) + .rejects.toThrow(/missing service\(s\) absent-service/) + await runtime.dispose() + }) + + it('runtime dispose is idempotent, unmounts views, disposes mounted features, and clears persisted state', async () => { + const runtime = await runtimeWithFrame() + const feature = await runtime.mount({ + inject: ['slots'], + apply: (ctx: typeof runtime.ctx) => { ctx.slots.register({ name: 'trt.panel' }, () => p) }, + }) + const view = runtime.renderRoot() + expect(view.container.textContent).toContain('p') + localStorage.setItem('trt.leftover', 'x') + + await runtime.dispose() + expect(view.container.innerHTML).toBe('') + expect(feature.fiber.uid).toBeNull() + expect(localStorage.getItem('trt.leftover')).toBeNull() + await runtime.dispose() // idempotent + await expect(runtime.dispose()).resolves.toBeUndefined() + }) +}) + +describe('single-slot mounting (declare + renderSlot)', () => { + it('renders one slot inside its data-slot wrapper and updates owner props in place', async () => { + const runtime = await SlotTestRuntime.create() + await runtime.declare({ 'trt.panel': { kind: 'single', scope: 'root' } }) + runtime.slots.register( + { name: 'trt.panel' }, + ({ label }: { label?: string }) => {label ?? 'none'}) + const slot = runtime.renderSlot('trt.panel', { label: 'first' }) + expect(slot.container.getAttribute('data-slot')).toBe('trt.panel') + expect(slot.view.getByTestId('panel').textContent).toBe('first') + + const panel = slot.view.getByTestId('panel') + slot.update({ label: 'second' }) + expect(slot.view.getByTestId('panel').textContent).toBe('second') + // In-place re-render: the element identity survived the owner flip. + expect(slot.view.getByTestId('panel')).toBe(panel) + await runtime.dispose() + }) + + it('views sibling slots of one tree separately and rejects undeclared keys', async () => { + const runtime = await SlotTestRuntime.create() + await runtime.declare({ + 'trt.panel': { kind: 'single', scope: 'root' }, + 'trt.rows': { kind: 'list', scope: 'root' }, + }) + runtime.slots.register({ name: 'trt.panel' }, () => panel) + runtime.slots.register({ name: 'trt.rows', id: 'r1' }, () => row) + const panel = runtime.renderSlot('trt.panel', {}) + const rows = runtime.renderSlot('trt.rows', {}) + expect(panel.container.textContent).toBe('panel') + expect(rows.container.textContent).toBe('row') + expect(() => runtime.renderSlot('trt.chat', {})).toThrow(/without declare\(\)/) + await runtime.dispose() + }) + + it('folds class hashes and collapses svg internals in snapshots, leaving the live DOM alone', async () => { + const runtime = await SlotTestRuntime.create() + await runtime.declare({ 'trt.panel': { kind: 'single', scope: 'root' } }) + runtime.slots.register({ name: 'trt.panel' }, () => ( +
+ styled + +
+ )) + const slot = runtime.renderSlot('trt.panel', {}) + expect(slot.container).toMatchSnapshot() + // The serializer works on a clone: the live DOM keeps hashes and paths. + expect(slot.container.querySelector('div')!.className).toBe('_frame_a1b2c3 plain') + expect(slot.container.querySelector('svg path')).not.toBeNull() + await runtime.dispose() + }) +}) + +describe('fixture session face', () => { + it('fail-loud stubs name the missing verb; supplied overrides run instead', async () => { + const runtime = await SlotTestRuntime.create() + await runtime.sessions.add({ id: 's1' }) + const bare = runtime.sessions.behavior('s1') + expect(() => bare.prompt()).toThrow(/prompt is not stubbed/) + expect(() => bare.cancel()).toThrow(/cancel is not stubbed/) + expect(() => bare.loadOlder()).toThrow(/loadOlder is not stubbed/) + expect(() => bare.loadAllHistory()).toThrow(/loadAllHistory is not stubbed/) + await runtime.dispose() + }) + + it('projections faces are identity-stable per key, read absent, and notify on set', async () => { + const runtime = await SlotTestRuntime.create() + await runtime.sessions.add({ id: 's1' }) + const session = runtime.sessions.behavior('s1') + const face = session.projections.faceOf('todos') + expect(session.projections.faceOf('todos')).toBe(face) + expect(face.getSnapshot()).toBeUndefined() + const seen: unknown[] = [] + const off = face.subscribe(() => { seen.push(face.getSnapshot()) }) + session.projections.set('todos', [1, 2]) + expect(seen).toEqual([[1, 2]]) + off() + session.projections.set('todos', [3]) + expect(seen).toEqual([[1, 2]]) // unsubscribed + // A never-subscribed key sets without listeners (the empty-notify arm). + session.projections.set('untouched', 1) + // The provide bundle hands the same store to the render side. + const info = runtime.sessions.provideInfo('s1')! + expect(info.projections?.faceOf('todos').getSnapshot()).toEqual([3]) + // A roster change rebuilds the ALREADY-materialized bundle eagerly + // (production channel semantics: mounted entries must see the provider) + // and skips never-materialized records (they pick the roster up lazily). + await runtime.sessions.add({ id: 's-lazy' }, { current: false }) + const offProbe = runtime.sessions.provide({ + hooks: ['probe2'], + resolve: () => ({ hooks: { probe2: { getSnapshot: () => 1, subscribe: () => () => {} } } }), + }) + const rebuilt = runtime.sessions.provideInfo('s1')! + expect(rebuilt).not.toBe(info) + expect(rebuilt.hooks['probe2']).toBeDefined() + offProbe() + await runtime.dispose() + }) +}) + +describe('workspaces action face', () => { + it('records every IWorkspaces verb with inert defaults and honors stubs', async () => { + const runtime = await SlotTestRuntime.create() + const ws = runtime.workspaces + const created = await ws.create({ name: 'alpha' }) + expect(created.title).toBe('alpha') + const registered = await ws.create({ path: '/tmp/beta' }) + expect(registered.path).toBe('/tmp/beta') + await expect(ws.pickDirectory()).resolves.toBeNull() + const renamed = await ws.rename('w1' as WorkspaceId, 'Renamed') + expect(renamed.title).toBe('Renamed') + await ws.delete('w1' as WorkspaceId) + await ws.openPath('/proj/file.ts') + const moved = await ws.insertSessionBefore('w1' as WorkspaceId, 's1' as SessionId, 's2' as SessionId) + expect(moved.sessionIds).toEqual(['s1']) + expect(ws.calls.map(c => c.method)).toEqual( + ['create', 'create', 'pickDirectory', 'rename', 'delete', 'openPath', 'insertSessionBefore']) + + ws.stub('create', () => Promise.resolve({ workspaceId: 'ws-x', title: 'X', path: '/x', sessionIds: [] } as never)) + ws.stub('pickDirectory', () => Promise.resolve('/picked')) + ws.stub('rename', () => Promise.resolve({ workspaceId: 'w1', title: 'S', path: '/s', sessionIds: [] } as never)) + ws.stub('delete', () => Promise.resolve()) + ws.stub('openPath', () => Promise.resolve()) + ws.stub('insertSessionBefore', () => Promise.resolve({ workspaceId: 'w1', title: '', path: '', sessionIds: [] } as never)) + expect((await ws.create({ name: 'y' })).title).toBe('X') + await expect(ws.pickDirectory()).resolves.toBe('/picked') + expect((await ws.rename('w1' as WorkspaceId, 'z')).title).toBe('S') + await ws.delete('w1' as WorkspaceId) + await ws.openPath('/other') + expect((await ws.insertSessionBefore('w1' as WorkspaceId, 's1' as SessionId)).sessionIds).toEqual([]) + await runtime.dispose() + }) +}) + +describe('single-slot mounting edge arms', () => { + it('renderSlot fails loud after dispose and after an external unmount', async () => { + const runtime = await SlotTestRuntime.create() + await runtime.declare({ 'trt.panel': { kind: 'single', scope: 'root' } }) + runtime.slots.register({ name: 'trt.panel' }, () => p) + runtime.renderSlot('trt.panel', {}) + // RTL cleanup empties the mounted tree behind the runtime's back: the + // wrapper lookup names the state instead of returning a dead container. + cleanup() + expect(() => runtime.renderSlot('trt.panel', {})).toThrow(/rendered no wrapper/) + await runtime.dispose() + // After dispose the root registration is gone: the production boot-order + // check fires before any wrapper lookup. + expect(() => runtime.renderSlot('trt.panel', {})).toThrow(/'root' has no registration/) + }) + + it('serializes childless svg untouched next to scoped classes', async () => { + const runtime = await SlotTestRuntime.create() + await runtime.declare({ 'trt.panel': { kind: 'single', scope: 'root' } }) + runtime.slots.register({ name: 'trt.panel' }, () => ( +
+ +
+ )) + const slot = runtime.renderSlot('trt.panel', {}) + expect(slot.container).toMatchSnapshot() + await runtime.dispose() + }) +}) diff --git a/packages/client/test-runtime/tsconfig.json b/packages/client/test-runtime/tsconfig.json new file mode 100644 index 0000000000..3e8a8561f8 --- /dev/null +++ b/packages/client/test-runtime/tsconfig.json @@ -0,0 +1,27 @@ +{ + "extends": "../../../tsconfig.base.client.json", + "compilerOptions": { + "rootDir": "src", + "outDir": "lib/types" + }, + "include": [ + "src" + ], + "references": [ + { + "path": "../../../vendor/cordis" + }, + { + "path": "../ui-slots" + }, + { + "path": "../web-react" + }, + { + "path": "../runtime" + }, + { + "path": "../../support/invariants" + } + ] +} diff --git a/packages/client/ui-command/src/client/service.ts b/packages/client/ui-command/src/client/service.ts index 22f4a911b3..8f9f96731f 100644 --- a/packages/client/ui-command/src/client/service.ts +++ b/packages/client/ui-command/src/client/service.ts @@ -9,10 +9,10 @@ import { Service } from 'cordis' import type { Context } from 'cordis' import type { ConnectionHandle, SessionId } from '@deepseek-ai/dsh-client-connection/client' -import type { ClientContext, SessionsService } from '@deepseek-ai/dsh-client-runtime/client' +import type { ClientContext, ISessions } from '@deepseek-ai/dsh-client-runtime/client' import type { CandidateRequest, ClientSessionContext, CommandClaim, PickOutcome, SlashCandidate, SlashPick, - SlashServiceContract, SubmitOutcome, + SubmitOutcome, } from '@deepseek-ai/dsh-client-ui-slash/client' import type { CommandContribution, CommandServiceContract } from './contract.ts' import type { CommandDescriptor } from './directory.ts' @@ -46,7 +46,7 @@ export class CommandService extends Service implements CommandServiceContract { if (!result.ok) throw new Error(`command.list failed: ${result.error.code}: ${result.error.message}`) return result.value.commands }) - const slash = ctx.get('slash') as SlashServiceContract | undefined + const slash = ctx.get('slash') if (slash === undefined) throw new Error('ui-command: slash service unavailable') ctx.effect(() => slash.registerSource({ trigger: '/', @@ -292,7 +292,7 @@ export class CommandService extends Service implements CommandServiceContract { return this.sessions().scope(id) } - private sessions(): SessionsService { + private sessions(): ISessions { const sessions = this.ctx.get('sessions') if (sessions === undefined) throw new Error('ui-command: sessions service unavailable') return sessions diff --git a/packages/client/ui-conversation/README.i18n.yaml b/packages/client/ui-conversation/README.i18n.yaml index b76d2f5447..e38a3f469d 100644 --- a/packages/client/ui-conversation/README.i18n.yaml +++ b/packages/client/ui-conversation/README.i18n.yaml @@ -2,5 +2,5 @@ # side as of the last confirmed-consistent state. Both languages carry equal authority; # after editing either side, bring the other along and re-record with: # pnpm run verify-translation-pairing --write packages/client/ui-conversation/README.md -README.md: b5067a0a4ad98b784439fa16211b6269fc44139d -README.zh.md: 5a8ba493c25f38d8be366d29462a258d8ba79ee9 +README.md: 7ab9cd3a065d6dd87c47dfc7056738351813a889 +README.zh.md: e38cb9ab837212546dd9680d5f2bf37865cb7e28 diff --git a/packages/client/ui-conversation/README.md b/packages/client/ui-conversation/README.md index b5067a0a4a..7ab9cd3a06 100644 --- a/packages/client/ui-conversation/README.md +++ b/packages/client/ui-conversation/README.md @@ -16,7 +16,7 @@ The todo surfaces are two registrations over that shape, both plain registrant p Per-session UI state for selection and the active view lives in the declared chat store (`stores.ts` `createChatStore`); the InputHub owns the composer state machine and mirrors its draft into that store for persistence. Apply passes one store handle to the strict session subtree, chat view, and details registrations, so each session shares one instance and the framework owns its lifecycle. Components are pure: the framework standard kit supplies `useSession`/`sessionId`, global `useSessions`/`useWorkspaces`, and the input machine's `useInput`/`inputActions`; store faces and inject factories supply the remaining state and callbacks. -The composer bar declares session-scoped single seats for `'conversation.input.plan'` and `'conversation.input.model'`, plus list slots for overlay, dock, left, and right input extensions. InputBar renders the model seat immediately before its pending indicator and send/stop button. Feature packages own each control and its state; ui-conversation supplies placement, the `locked` owner prop, and the standard slot shares. A pending composer takeover remains mounted when another conversation view is active so the blocked agent can still receive its answer; without a pending interaction, the active-session composer belongs to Chat. The resident no-session shell uses `DisabledInputBar` and therefore dispatches no session-scoped control seats. +The composer bar declares session-scoped single seats for `'conversation.input.plan'` (right of the local access-mode control) and `'conversation.input.model'` (immediately before the pending indicator and send/stop button), plus list slots for overlay, dock, left, and right input extensions. Feature packages own each control and its state; ui-conversation supplies placement, the `locked` owner prop, and the standard slot shares. While the `plan` projection's effective target is plan mode, InputBar swaps its textarea placeholder to the plan-task wording (a host-folded value read through the standard-kit `useProjection`; owner-supplied placeholders win). A pending composer takeover remains mounted when another conversation view is active so the blocked agent can still receive its answer; without a pending interaction, the active-session composer belongs to Chat. The resident no-session shell uses `DisabledInputBar` and therefore dispatches no session-scoped control seats. `src/client/` is organized for the future package split: `contract/` is the sole inter-domain shared face (`slots.ts` slot declarations + composed slot props including the tool-row contract, `views.ts` shared primitives, `tool-call-model.ts`); the `skeleton/`, `chat/`, and `toolviews/` (sample registrants) domain directories import contract files and never each other; `apply.ts` is the only assembly point allowed to import all three domains. The `/client` export surface is the contract only — `apply`/`inject`, the two service classes, and the `contract/` type families; implementation components (skeleton, chat rows) and the store factory stay internal and reach the page exclusively through apply's slot registrations (tests take them via the `./src/*` subpath). diff --git a/packages/client/ui-conversation/README.zh.md b/packages/client/ui-conversation/README.zh.md index 5a8ba493c2..e38cb9ab83 100644 --- a/packages/client/ui-conversation/README.zh.md +++ b/packages/client/ui-conversation/README.zh.md @@ -16,7 +16,7 @@ todo 两个面就是在该形状上的两个注册项，都是普通注册方插 逐 Session UI 状态中的选择与活跃视图位于已声明的聊天 store（`stores.ts` `createChatStore`）中；InputHub 拥有输入区状态机，并将草稿镜像到该 store 以便持久化。apply 将同一个 store handle 传给严格限定于会话的子树、聊天视图和详情注册，因此每个会话内共享一个实例，框架拥有其生命周期。组件保持纯粹：框架标准工具包提供 `useSession`／`sessionId`、全局 `useSessions`／`useWorkspaces`，以及输入状态机的 `useInput`／`inputActions`；store 表层与 inject factory 提供其余状态和回调。 -输入栏为 `'conversation.input.plan'` 和 `'conversation.input.model'` 声明会话作用域的单实例 seat，并为 overlay、dock、left 和 right 输入扩展声明列表 slot。InputBar 将模型 seat 渲染在 pending 指示器与发送／停止按钮之前。各功能包拥有相应控件及其状态；ui-conversation 提供放置位置、`locked` owner prop 和标准 slot share。另一个会话视图活跃时，待处理的 composer 接管仍保持挂载，使被阻塞的 agent（智能体）仍能收到回答；没有待处理交互时，活跃会话的 composer 归 Chat 所有。常驻无会话壳使用 `DisabledInputBar`，因此不会分发任何会话作用域的控件 seat。 +输入栏声明两个会话作用域的单实例 seat：`'conversation.input.plan'` 位于本地 access 模式控件右侧，而 `'conversation.input.model'` 紧接在 pending 指示器与发送／停止按钮之前；它还为 overlay、dock、left 和 right 输入扩展声明列表 slot。各功能包拥有相应控件及其状态；ui-conversation 提供放置位置、`locked` owner prop 和标准 slot share。当 `plan` 投影的有效目标为 plan mode 时，InputBar 会将文本框的 placeholder 切换为 plan 任务文案（它通过标准工具包的 `useProjection` 读取由 host 折叠的值；owner 提供的 placeholder 优先）。另一个会话视图活跃时，待处理的 composer 接管仍保持挂载，使被阻塞的 agent（智能体）仍能收到回答；没有待处理交互时，活跃会话的 composer 归 Chat 所有。常驻无会话壳使用 `DisabledInputBar`，因此不会分发任何会话作用域的控件 seat。 `src/client/` 按未来的包拆分组织：`contract/` 是唯一的跨领域共享表层（`slots.ts` slot 声明 + 组合后的 slot props，包括工具行契约、`views.ts` 共享原语、`tool-call-model.ts`）；`skeleton/`、`chat/` 和 `toolviews/`（示例注册方）领域目录只导入 contract 文件，彼此绝不导入；`apply.ts` 是唯一允许导入全部三个领域的组装点。`/client` 导出表层只包含契约：`apply`／`inject`、两个服务类和 `contract/` 类型家族；实现组件（骨架、聊天行）与 store factory 保持内部状态，只能通过 apply 的 slot 注册到达页面（测试通过 `./src/*` 子路径获取它们）。 diff --git a/packages/client/ui-conversation/package.json b/packages/client/ui-conversation/package.json index 0b184dcbdf..2e37415bdc 100644 --- a/packages/client/ui-conversation/package.json +++ b/packages/client/ui-conversation/package.json @@ -49,6 +49,7 @@ }, "devDependencies": { "@deepseek-ai/dsh-client-runtime": "workspace:^", + "@deepseek-ai/dsh-plan-mode": "workspace:^", "@deepseek-ai/dsh-session-projection": "workspace:^", "@deepseek-ai/dsh-tool-todo": "workspace:^", "@deepseek-ai/dsh-client-ui-layout": "workspace:^", diff --git a/packages/client/ui-conversation/src/client/apply.ts b/packages/client/ui-conversation/src/client/apply.ts index b6a971bd9f..e56048c2db 100644 --- a/packages/client/ui-conversation/src/client/apply.ts +++ b/packages/client/ui-conversation/src/client/apply.ts @@ -1,7 +1,7 @@ /** Registers the conversation components, shared store, and service callbacks. */ import type { Context } from 'cordis' import type { BoundActions } from '@deepseek-ai/dsh-client-ui-slots' -import type { SessionId, SessionsService } from '@deepseek-ai/dsh-client-runtime/client' +import type { ISessions, SessionId } from '@deepseek-ai/dsh-client-runtime/client' import type {} from '@deepseek-ai/dsh-client-ui-layout/client' import type { ViewTab } from './contract/views.ts' import type { @@ -10,6 +10,7 @@ import type { import { resolveToolPath } from './contract/tool-call-model.ts' import { createChatStore } from './stores.ts' import { ConversationService } from './service.ts' +import type { IConversation } from './service.ts' import { InputHub } from './input/hub.ts' import { InputBar } from './skeleton/InputBar.tsx' import { ChatView } from './chat/ChatView.tsx' @@ -24,8 +25,8 @@ import { DetailsPanel } from './skeleton/DetailsPanel.tsx' /** Services required by the conversation plugin. */ export const inject = ['slots', 'layout', 'sessions', 'workspaces'] -/** Resolve the session-scoped conversation service (scope-addressed send/cancel), failing loud. */ -function scopedConversation(sessions: SessionsService, id: SessionId): ConversationService { +/** Resolve the session-scoped conversation face (scope-addressed send/cancel), failing loud. */ +function scopedConversation(sessions: ISessions, id: SessionId): IConversation { const scoped = sessions.scope(id) if (scoped === undefined) throw new Error(`ui-conversation: session "${id}" resolved no scope`) const conversation = scoped.get('conversation') @@ -129,8 +130,9 @@ export function apply(ctx: Context): void { // verbs ride this inject (package-internal — hub and bar are one plugin). slots.register({ name: 'conversation.composer.bar', - // The two named control seats in the bar's tool row (plan left, model - // right); empty until their owning plugins register (B ruling). + // The two named control seats in the bar's tool row (plan beside the + // access control, model right); empty until their owning plugins + // register (B ruling). children: { 'conversation.input.plan': { kind: 'single', scope: 'session' }, 'conversation.input.model': { kind: 'single', scope: 'session' }, diff --git a/packages/client/ui-conversation/src/client/chat/ChatView.tsx b/packages/client/ui-conversation/src/client/chat/ChatView.tsx index 5273c75357..e9503b5677 100644 --- a/packages/client/ui-conversation/src/client/chat/ChatView.tsx +++ b/packages/client/ui-conversation/src/client/chat/ChatView.tsx @@ -375,7 +375,9 @@ export function ChatView({ useSession, useSessions, useStore, renderSlot, sessio ))}
)} - {pending.map(item => )} + {pending.map(item => item.kind === 'approval' + ? + : null)} {/* Turn-level loading signal: rides the whole running turn (first-token wait, tool execution, streaming) so it never flickers per step. */} {running && } diff --git a/packages/client/ui-conversation/src/client/chat/GenericToolCard.tsx b/packages/client/ui-conversation/src/client/chat/GenericToolCard.tsx index 7bd4f22b06..412ff9fe59 100644 --- a/packages/client/ui-conversation/src/client/chat/GenericToolCard.tsx +++ b/packages/client/ui-conversation/src/client/chat/GenericToolCard.tsx @@ -6,12 +6,12 @@ import type { ReactNode } from 'react' import { - IconApiOutline14, IconBrowseOutline16, IconCodeOutline16, IconEditOutline16, IconSearchOutline16, IconThinkOutline14, + IconApiOutline14, IconBrowseOutline16, IconCodeOutline16, IconEditOutline16, IconSearchOutline16, IconSparkle16, + IconThinkOutline14, } from '@deepseek-ai/dsh-client-ui-primitives' import type { ToolRowOwnerProps } from '../contract/slots.ts' import { toolRowModel, type ToolRowVariant } from '../contract/tool-call-model.ts' import { ToolRow } from './ToolRow.tsx' -import { IconSparkle16 } from './IconSparkle16.tsx' /** Variant leading icons (figma table); all glyphs render at 14 inside the 16px leading box. */ const VARIANT_ICONS: Record = { diff --git a/packages/client/ui-conversation/src/client/chat/IconSparkle16.tsx b/packages/client/ui-conversation/src/client/chat/IconSparkle16.tsx deleted file mode 100644 index 61331ba616..0000000000 --- a/packages/client/ui-conversation/src/client/chat/IconSparkle16.tsx +++ /dev/null @@ -1,15 +0,0 @@ -// Local sparkle icon for the Others tool-row variant (figma 43:31850 leading -// glyph is an SF Symbols "sparkles" text glyph — not extractable as vector -// data, so this is a hand-authored three-star approximation). Lives here -// rather than ui-primitives until the exact glyph is exported and adopted -// into the ic_ds_* family. - -export function IconSparkle16({ size = 16, className }: { size?: number; className?: string }) { - return ( - - - - - - ) -} diff --git a/packages/client/ui-conversation/src/client/chat/PendingCard.tsx b/packages/client/ui-conversation/src/client/chat/PendingCard.tsx index b6825aed9a..ee72252717 100644 --- a/packages/client/ui-conversation/src/client/chat/PendingCard.tsx +++ b/packages/client/ui-conversation/src/client/chat/PendingCard.tsx @@ -1,30 +1,19 @@ -// PendingCard: approval/question placeholder card (visible, not answerable — -// the composer-takeover approval panel is a P-II item; wire pending semantics -// already exist so the flow must show them). +// PendingCard: display-only approval placeholder. Questions render exclusively +// through the composer takeover so the same pending wait is never shown twice. import { memo } from 'react' -import type { PendingInteraction } from '@deepseek-ai/dsh-client-runtime/client' -import { JsonBlock } from '@deepseek-ai/dsh-client-ui-primitives' +import type { PendingWait } from '@deepseek-ai/dsh-client-runtime/client' import css from './PendingCard.module.css' export interface PendingCardProps { - item: PendingInteraction + item: PendingWait<'approval'> } export const PendingCard = memo(function PendingCard({ item }: PendingCardProps) { return (
- {item.kind === 'approval' ? ( - <> -
等待审批：{item.payload.toolName}
- {item.payload.reason !== undefined &&
{item.payload.reason}
} - - ) : ( - <> -
等待回答（{item.payload.questions.length} 题）
- - - )} +
等待审批：{item.payload.toolName}
+ {item.payload.reason !== undefined &&
{item.payload.reason}
}
请在原客户端处理（web 端作答后续里程碑提供）
) diff --git a/packages/client/ui-conversation/src/client/contract/slots.ts b/packages/client/ui-conversation/src/client/contract/slots.ts index 63b816546f..f756c7ce89 100644 --- a/packages/client/ui-conversation/src/client/contract/slots.ts +++ b/packages/client/ui-conversation/src/client/contract/slots.ts @@ -83,9 +83,10 @@ declare module '@deepseek-ai/dsh-client-ui-slots' { */ 'conversation.composer.bar': { kind: 'single'; scope: 'session'; owner: ComposerBarOwnerProps } /** - * The Plan-mode control seat in the composer tool row (left group). - * Declared by the composer-bar entry; empty until a plan plugin - * registers (B ruling: no placeholder fallback). + * The Plan-mode status seat in the composer tool row (left group, + * right of the access-mode control). Declared by the composer-bar + * entry; empty until a plan plugin registers (B ruling: no placeholder + * fallback). */ 'conversation.input.plan': { kind: 'single'; scope: 'session'; owner: InputControlOwnerProps } /** diff --git a/packages/client/ui-conversation/src/client/index.ts b/packages/client/ui-conversation/src/client/index.ts index 1b85c52abb..56d398def3 100644 --- a/packages/client/ui-conversation/src/client/index.ts +++ b/packages/client/ui-conversation/src/client/index.ts @@ -3,10 +3,9 @@ * between the independently implemented skeleton and chat domains; `apply.ts` * owns their slot assembly. */ -import type { ConversationService } from './service.ts' - export { apply, inject } from './apply.ts' export { ConversationService } from './service.ts' +export type { IConversation } from './service.ts' export type { CallId, ChatStoreState, SelectionTarget, ViewTab, @@ -22,6 +21,7 @@ export type { declare module 'cordis' { interface Context { - conversation: ConversationService + /** The outward face only; the concrete service stays inside this plugin. */ + conversation: import('./service.ts').IConversation } } diff --git a/packages/client/ui-conversation/src/client/input/hub.ts b/packages/client/ui-conversation/src/client/input/hub.ts index 2ae474be31..2641e0dcc4 100644 --- a/packages/client/ui-conversation/src/client/input/hub.ts +++ b/packages/client/ui-conversation/src/client/input/hub.ts @@ -8,9 +8,8 @@ * bail events) and owns the default-sink choreography: every session is a * real host entity, so the sink is one unconditional prompt path. */ -import type { ClientContext, Session, SessionBinding, SessionId, SessionsService } from '@deepseek-ai/dsh-client-runtime/client' -import type { SlashController, SlashServiceContract } from '@deepseek-ai/dsh-client-ui-slash/client' -import type {} from '@deepseek-ai/dsh-client-ui-slash/client' +import type { ClientContext, ISessions, SessionBinding, SessionFace, SessionId } from '@deepseek-ai/dsh-client-runtime/client' +import type { SlashController } from '@deepseek-ai/dsh-client-ui-slash/client' import { queueReadFaceOf } from '../queue/store.ts' import type { ComposerKeyboard, InputService, SessionInput } from './contract.ts' import type { PopupDismissFace } from './facade.ts' @@ -113,7 +112,7 @@ export class InputHub implements InputService { * exactly one path; a failed first prompt is an ordinary prompt failure * (error strip via promptError, draft restored only while untouched). */ - private sink(session: Session, text: string, mode: 'queue' | 'steer'): void { + private sink(session: SessionFace, text: string, mode: 'queue' | 'steer'): void { if (text === '') return const shell = this.shells.get(session.sessionId) // Commit, not an editable clear: undo must not resurrect sent content. @@ -129,7 +128,7 @@ export class InputHub implements InputService { } private controller(actx: ClientContext): SlashController | undefined { - const slash = this.rootCtx.get('slash') as SlashServiceContract | undefined + const slash = this.rootCtx.get('slash') return slash?.sessionOf(actx) } @@ -138,7 +137,7 @@ export class InputHub implements InputService { return command?.popupFor(actx) } - private sessions(): SessionsService { + private sessions(): ISessions { const sessions = this.rootCtx.get('sessions') if (sessions === undefined) throw new Error('conversation.input: sessions service unavailable') return sessions diff --git a/packages/client/ui-conversation/src/client/queue/store.ts b/packages/client/ui-conversation/src/client/queue/store.ts index 5d113b750d..6f084ea39d 100644 --- a/packages/client/ui-conversation/src/client/queue/store.ts +++ b/packages/client/ui-conversation/src/client/queue/store.ts @@ -5,7 +5,7 @@ * reference-stable across unrelated snapshot swaps, so this is a pure * projection — no second store, no copy. */ -import type { ObservableSnapshot, Session } from '@deepseek-ai/dsh-client-runtime/client' +import type { ObservableSnapshot, SessionFace } from '@deepseek-ai/dsh-client-runtime/client' import type { QueuedMessage } from '../input/contract.ts' /** @@ -13,10 +13,10 @@ import type { QueuedMessage } from '../input/contract.ts' * The wiring layer (T5) overlays this onto InputState.queue; the runtime * QueuedMessage and the input-contract QueuedMessage are structurally the * same frozen shape ({key, preview}). - * @param session - the resident session instance. + * @param session - the resident session face. * @returns the queue read face (snapshot reference stable while the queue is unchanged). */ -export function queueReadFaceOf(session: Session): ObservableSnapshot { +export function queueReadFaceOf(session: SessionFace): ObservableSnapshot { return { getSnapshot: () => session.getSnapshot().queue, subscribe: fn => session.subscribe(fn), diff --git a/packages/client/ui-conversation/src/client/service.ts b/packages/client/ui-conversation/src/client/service.ts index 0d2d8e9d8e..7b119b15ff 100644 --- a/packages/client/ui-conversation/src/client/service.ts +++ b/packages/client/ui-conversation/src/client/service.ts @@ -12,24 +12,50 @@ import type { Context } from 'cordis' // Type-only imports: a plugin-to-plugin value import is a bundle purity // error, so scope resolution goes through the sessions service (scopeOf // method) instead of the standalone helper. -import type { Session, SessionId, SessionsService } from '@deepseek-ai/dsh-client-runtime/client' -import { InputHub } from './input/hub.ts' +import type { ISessions, SessionFace, SessionId } from '@deepseek-ai/dsh-client-runtime/client' +import type { InputService } from './input/contract.ts' + +/** + * The outward conversation face (`ctx.conversation`): the scope-addressed + * verbs and the input registry other plugins may reach — and exactly what a + * test fake must supply. + */ +export interface IConversation { + /** The per-session input machine registry (InputService face). */ + readonly input: InputService + /** + * Send a prompt into the caller scope's session. + * @param text - prompt text, sent verbatim as one text block. + * @param mode - queue after the current turn, or steer into it. + * @returns completion; business failures reject (and land in promptError). + */ + send(text: string, mode: 'queue' | 'steer'): Promise + /** + * Cancel the scoped session's in-flight turn. + * @returns completion; failures reject as in send. + */ + cancel(): Promise + /** + * Pull one older history page for the scoped session. + * @returns completion of the page pull. + */ + loadOlder(): Promise +} /** Scope-addressed conversation service (root singleton, provided as `conversation`). */ -export class ConversationService extends Service { +export class ConversationService extends Service implements IConversation { /** The per-session input machine registry (InputService face, design §5.2). */ - readonly input: InputHub + readonly input: InputService /** * @param ctx - owning root context (the plugin apply context; the service * registers itself and follows that fiber's lifetime). - * @param config - the shared InputHub constructed by the plugin apply - * (shared with the slot inject factories); absent = own instance - * (object-layer tests that never touch slots). + * @param config - carries the InputService instance constructed by the + * plugin apply (the same InputHub the slot inject factories close over). */ - constructor(ctx: Context, config?: { input?: InputHub }) { + constructor(ctx: Context, config: { input: InputService }) { super(ctx, 'conversation') - this.input = config?.input ?? new InputHub(ctx) + this.input = config.input } /** @@ -57,8 +83,8 @@ export class ConversationService extends Service { await this.scopedSession('loadOlder').loadOlder() } - /** Resolve the caller scope's Session or throw on root contexts. */ - private scopedSession(op: string): Session { + /** Resolve the caller scope's session face or throw on root contexts. */ + private scopedSession(op: string): SessionFace { const id = this.scopeId(op) const binding = this.requireSessions().binding(id) if (binding === undefined) throw new Error(`conversation.${op}: session "${id}" resolved no binding`) @@ -74,10 +100,9 @@ export class ConversationService extends Service { return id } - private requireSessions(): SessionsService { - // ctx.get instead of ctx.sessions: the typed Context merge is suspended - // while the client/host `sessions` declaration collision awaits - // arbitration (see the runtime package's Context merge note). + private requireSessions(): ISessions { + // Strict ctx.get, not the injection proxy: the scope-addressed pattern + // reads the service off whatever context the tracker rebound. const sessions = this.ctx.get('sessions') if (sessions === undefined) throw new Error('conversation: sessions service unavailable') return sessions diff --git a/packages/client/ui-conversation/src/client/skeleton/InputBar.tsx b/packages/client/ui-conversation/src/client/skeleton/InputBar.tsx index 26a65def6a..c9476bd653 100644 --- a/packages/client/ui-conversation/src/client/skeleton/InputBar.tsx +++ b/packages/client/ui-conversation/src/client/skeleton/InputBar.tsx @@ -10,6 +10,9 @@ import { useEffect, useRef, useState } from 'react' import type { ChangeEvent, KeyboardEvent, MouseEvent, ReactNode } from 'react' import clsx from 'clsx' import { IconPlusOutline16 } from '@deepseek-ai/dsh-client-ui-primitives' +// Type-only: the `plan` projection key merge (the TodoDock posture — the +// composer reads a host-computed value; the domain owns the key). +import type {} from '@deepseek-ai/dsh-plan-mode/client' import type { ComposerBarProps } from '../contract/slots.ts' import { deriveDecorations } from '../input/decorations.ts' import css from './InputBar.module.css' @@ -28,7 +31,7 @@ const READONLY_OPTIONS: readonly { id: string; label: string }[] = [ ] export function InputBar({ - useSession, useInput, inputActions, keyboard, stop, renderSlot, useNotices, useLexicon, + useSession, useInput, inputActions, keyboard, stop, renderSlot, useNotices, useLexicon, useProjection, variant, placeholder, accessory, overlay, leftItems, rightItems, onAdd, addLabel = 'Add attachment', }: InputBarProps) { const input = useInput(s => s) @@ -37,6 +40,9 @@ export function InputBar({ const promptError = useSession(s => s.promptError) const running = useSession(s => s.running) const disabled = useSession(s => s.removed) + // Plan mode swaps the textarea placeholder (the projection is the folded + // host value; owner-prop placeholders — hero, session-unavailable — win). + const planActive = useProjection('plan', plan => plan !== undefined && (plan.pending ? !plan.active : plan.active)) // Prompt failures are ordinary failures (no create/attach transaction // exists anymore): the strip renders promptError, the draft stays in the // machine, and the user resubmits. @@ -334,7 +340,9 @@ export function InputBar({ disabled={locked} readOnly={machineBusy} data-phase={input.phase} - placeholder={placeholder ?? (disabled ? 'Session unavailable' : 'Message the agent')} + placeholder={placeholder ?? (disabled + ? 'Session unavailable' + : planActive ? 'describe your task to generate plan' : 'Message the agent')} rows={2} onChange={onChange} onKeyDown={onKeyDown} @@ -361,15 +369,15 @@ export function InputBar({
- {renderSlot('conversation.input.plan', { locked })} {accessSelect} + {renderSlot('conversation.input.plan', { locked })}
{leftItems}
{rightItems} {renderSlot('conversation.input.model', { locked })} - {machineBusy && } + {/* {machineBusy && } */} )) - const view = mountApp(b.slots) + const view = b.runtime.renderRoot() const row = view.getByTestId('probe-row') expect(row.textContent).toBe(`for:${SID}`) row.click() expect(poked).toEqual([SID]) + await b.runtime.dispose() }) }) describe('registrant load-order seam', () => { it("suspends a registrant on inject: ['slots', 'conversation'] until the service (and the hole) exists", async () => { - const ctx = new Context() - const slotsFiber = ctx.plugin(SlotsService) - await slotsFiber.await() - const slots = ctx.get('slots') as SlotsService - ctx.provide('sessions', { - list: createSnapshotStore({ - ids: [], byId: {}, current: undefined, phase: 'ready', - }), - binding: () => undefined, - scope: () => undefined, - provideInfo: () => undefined, - currentProvideInfo: { - getSnapshot: () => ABSENT_INFO, - subscribe: () => () => {}, - }, - provide: () => () => {}, - create: vi.fn(), - open: vi.fn(), - updateIntent: vi.fn(), - }) - ctx.provide('workspaces', { - list: createSnapshotStore({ - items: [], state: 'idle', phase: 'ready', error: null, - baselinesReady: true, recentWorkspaceId: undefined, - }), - startSession: vi.fn(), - sendSession: vi.fn(), - openPath: vi.fn(async () => {}), - }) - ctx.provide('layout', { openDetails: vi.fn(), closeDetails: vi.fn() }) - ctx.provide('locale', { bind: () => (key: string) => key }) - slots.register({ - name: 'root', - children: { - 'conversation': { kind: 'single', scope: 'session-maybe' }, - 'details': { kind: 'single', scope: 'session' }, - }, - }, AppRoot) + const runtime = await SlotTestRuntime.create() + runtime.provide('layout', { openDetails: vi.fn(), closeDetails: vi.fn() }) + await runtime.root.declare(LAYOUT_CHILDREN, AppRoot) // Third-party posture, mounted BEFORE ui-conversation: real fiber inject // semantics hold it — apply must not run while 'conversation' is absent. - // (Plain arrow, not vi.fn: mock functions carry a prototype and trip the - // fiber's isConstructor branch.) + // Uses ctx.plugin directly (the deliberate-suspension escape hatch; mount() + // would fail loud on the missing service). (Plain arrow, not vi.fn: mock + // functions carry a prototype and trip the fiber's isConstructor branch.) let applyRuns = 0 - const registrantApply = (registrantCtx: Context): void => { + const registrantApply = (registrantCtx: typeof runtime.ctx): void => { applyRuns += 1 registrantCtx.slots.register( { name: 'conversation.chat.toolview', key: 'late' }, () => null) } - const late = ctx.plugin({ + const late = runtime.ctx.plugin({ name: 'late-registrant', inject: ['slots', 'conversation'], apply: registrantApply, @@ -317,11 +205,11 @@ describe('registrant load-order seam', () => { // Mounting the package resolves the seam: service present ⟹ the chat // entry (and its hole declaration) is already on the ledger, so the // suspended registrant lands without an undeclared-slot throw. - const fiber = ctx.plugin({ inject: [...inject], apply }) - await fiber.await() + await runtime.mount({ inject: [...inject], apply }) await late.await() expect(applyRuns).toBe(1) - expect(slots.entries('conversation.chat.toolview').map(e => e.options.key)) + expect(runtime.slots.entries('conversation.chat.toolview').map(e => e.options.key)) .toEqual(expect.arrayContaining(['bash', 'late'])) + await runtime.dispose() }) }) diff --git a/packages/client/ui-conversation/tests/coverage-tails.spec.tsx b/packages/client/ui-conversation/tests/coverage-tails.spec.tsx index 90628f64ee..e2c5244ada 100644 --- a/packages/client/ui-conversation/tests/coverage-tails.spec.tsx +++ b/packages/client/ui-conversation/tests/coverage-tails.spec.tsx @@ -1,6 +1,6 @@ // @vitest-environment jsdom // Branch tails the acceptance specs do not reach: ToolRow stopped-state dot, -// PendingCard question arm, bash sample state dots, the node-half empty +// PendingCard approval wait, bash sample state dots, the node-half empty // apply, and AssistantMarkdown reasoning/unknown block arms. import { afterEach, describe, expect, it, vi } from 'vitest' @@ -33,11 +33,11 @@ describe('tails', () => { expect(view.container.querySelector('[data-state="stopped"]')).not.toBeNull() }) - it('PendingCard renders the question arm with its count', () => { + it('PendingCard renders the approval wait with its tool name', () => { const view = render( - ['payload'], vi.fn())} />, + ['payload'], vi.fn())} />, ) - expect(view.getByText(/等待回答（2 题）/)).toBeTruthy() + expect(view.getByText(/等待审批/)).toBeTruthy() }) it('AssistantMarkdown renders reasoning as a Think row and unknown blocks as JSON fallback', () => { diff --git a/packages/client/ui-conversation/tests/input-bar.spec.tsx b/packages/client/ui-conversation/tests/input-bar.spec.tsx index f8b9fd3d67..dbd9b180bd 100644 --- a/packages/client/ui-conversation/tests/input-bar.spec.tsx +++ b/packages/client/ui-conversation/tests/input-bar.spec.tsx @@ -30,6 +30,8 @@ function snapshotOf(overrides: Partial = {}): Conversation interface BenchOptions { planEntry?: React.ReactNode + /** The `plan` projection value the standard-kit useProjection serves. */ + plan?: { active: boolean; pending: boolean } modelEntry?: React.ReactNode /** Hot text-ref lexicon (injects a minimal slash stub exposing only lexicon()). */ lexicon?: ReadonlyMap<'/' | '@', readonly string[]> @@ -88,7 +90,8 @@ function bench(over?: BenchOptions) { items: [], state: 'idle', phase: 'ready', error: null, baselinesReady: true, recentWorkspaceId: undefined, })), - useProjection: (() => undefined), + useProjection: ((_key: string, selector?: (v: unknown) => unknown) => + (selector ?? (v => v))(over?.plan)), useInput: bindSnapshotSelector(shell.state), inputActions: shell.actions, keyboard: shell, @@ -230,6 +233,20 @@ describe('running and lock semantics (queue cut 1)', () => { const custom = bench({ placeholder: 'Custom placeholder' }) expect(custom.textarea.placeholder).toBe('Custom placeholder') }) + + it('the plan projection swaps the placeholder while its effective target is plan mode', () => { + const active = bench({ plan: { active: true, pending: false } }) + expect(active.textarea.placeholder).toBe('describe your task to generate plan') + // /plan just ran: pending entry already reads as the plan target. + const entering = bench({ plan: { active: false, pending: true } }) + expect(entering.textarea.placeholder).toBe('describe your task to generate plan') + // Pending exit: target is default again. + const leaving = bench({ plan: { active: true, pending: true } }) + expect(leaving.textarea.placeholder).toBe('Message the agent') + // Owner placeholder outranks the plan swap. + const custom = bench({ plan: { active: true, pending: false }, placeholder: 'Custom placeholder' }) + expect(custom.textarea.placeholder).toBe('Custom placeholder') + }) }) describe('machine pending lock', () => { @@ -250,7 +267,6 @@ describe('machine pending lock', () => { expect(shell.snapshot.phase).toBe('submitting') const textarea = view.container.querySelector('textarea')! expect(textarea.readOnly).toBe(true) - expect(view.container.querySelector('[data-input-pending]')).not.toBeNull() expect(view.container.querySelector('button[aria-label="Send message"]')!.disabled).toBe(true) }) }) diff --git a/packages/client/ui-conversation/tests/input-matrix.spec.tsx b/packages/client/ui-conversation/tests/input-matrix.spec.tsx index 8a04dde4a5..29769afd5d 100644 --- a/packages/client/ui-conversation/tests/input-matrix.spec.tsx +++ b/packages/client/ui-conversation/tests/input-matrix.spec.tsx @@ -124,13 +124,12 @@ describe('matrix row: claimed', () => { describe('matrix row: submitting', () => { it('locks enter, renders pending + read-only, keeps the claim snapshot on the currency', async () => { const submit = vi.fn(() => new Promise(() => {})) // never settles - const { view, textarea, shell, sink, claim } = bench({ submit }) + const { textarea, shell, sink, claim } = bench({ submit }) claim() fireEvent.keyDown(textarea, { key: 'Enter' }) expect(shell.snapshot.phase).toBe('submitting') expect(shell.snapshot.claim).toBeDefined() expect((textarea).readOnly).toBe(true) - expect(view.container.querySelector('[data-input-pending]')).not.toBeNull() // Enter is dead inside the lock (submit dispatch is microtask-deferred). await vi.waitFor(() => { expect(submit).toHaveBeenCalledTimes(1) }) fireEvent.keyDown(textarea, { key: 'Enter' }) diff --git a/packages/client/ui-conversation/tests/selection-survival.spec.ts b/packages/client/ui-conversation/tests/selection-survival.spec.ts deleted file mode 100644 index 533d5ce730..0000000000 --- a/packages/client/ui-conversation/tests/selection-survival.spec.ts +++ /dev/null @@ -1,149 +0,0 @@ -// @vitest-environment jsdom -/** - * Exercises selection persistence through the real SlotsService store axis; - * component stubs cannot prove per-session identity or disposal. - */ -import { Context } from 'cordis' -import { beforeEach, describe, expect, it } from 'vitest' -import { createSnapshotStore, SlotsService } from '@deepseek-ai/dsh-client-runtime/client' -import type { SessionId, SessionListState, WorkspaceListState } from '@deepseek-ai/dsh-client-runtime/client' -import { createChatStore } from '../src/client/stores.ts' - -const sid = (s: string): SessionId => s as SessionId - -/** Identity-stable no-session bundle (uSES getSnapshot contract). */ -const ABSENT_INFO = { sessionId: undefined, hooks: {}, props: {} } - -interface Bench { - slots: SlotsService - chat: ReturnType -} - -function bench(): Bench { - const ctx = new Context() - ctx.provide('sessions', { - list: createSnapshotStore({ - ids: [], byId: {}, current: undefined, phase: 'ready', - }), - provideInfo: () => undefined, - currentProvideInfo: { - getSnapshot: () => ABSENT_INFO, - subscribe: () => () => {}, - }, - provide: () => () => {}, - }) - ctx.provide('workspaces', { - list: createSnapshotStore({ - items: [], state: 'idle', phase: 'ready', error: null, - baselinesReady: true, recentWorkspaceId: undefined, - }), - }) - // Service self-registers as ctx 'slots' (cordis Service constructor). - const slots = new SlotsService(ctx) - const chat = createChatStore() - // The apply.ts shape: one shared handle across both session-slot - // registrations. 'conversation'/'details' must first exist in the ledger — - // register a root occupant declaring them (the AppFrame role; the stand-in - // consumes renderSlot to satisfy the declare-means-render check). - slots.register({ - name: 'root', - children: { - 'conversation': { kind: 'single', scope: 'session-maybe' }, - 'conversation.session': { kind: 'single', scope: 'session' }, - 'details': { kind: 'single', scope: 'session' }, - }, - }, (_p: { renderSlot?: unknown }) => null) - // apply.ts mounts the shared chat handle only under session-scope slots - // (the session-maybe 'conversation' shell carries no store). - slots.register({ name: 'conversation.session', store: chat }, () => null) - slots.register({ name: 'details', store: chat }, () => null) - return { slots, chat } -} - -/** Resolve the store instance the renderer would hand a slot's component for a session. */ -function storeFor(b: Bench, slot: 'conversation.session' | 'details', sessionId: SessionId) { - const host = renderHost(b) - const entry = host.entriesOf(slot)[0]! - return host.storeOf(entry, sessionId)! as ReturnType['create']> -} - -/** The host face is only built at renderSlot time; install a stub renderer once to reach it. */ -function renderHost(b: Bench): import('@deepseek-ai/dsh-client-ui-slots').SlotRendererHost { - const captured = (b as unknown as { _host?: import('@deepseek-ai/dsh-client-ui-slots').SlotRendererHost }) - if (captured._host === undefined) { - b.slots.install({ - renderRoot: (host) => { - captured._host = host - return null - }, - }) - b.slots.renderSlot('root', {}) - } - return captured._host! -} - -beforeEach(() => { - localStorage.clear() -}) - -describe('selection survives on the store seat', () => { - it('one session, two slots: conversation writes, details reads the SAME instance', () => { - const b = bench() - - const conv = storeFor(b, 'conversation.session', sid('s1')) - const details = storeFor(b, 'details', sid('s1')) - conv.actions.select({ turnSeq: 3, callId: 'c1' }) - expect(details.store.getSnapshot().selection).toEqual({ turnSeq: 3, callId: 'c1' }) - // Identity, not just value: the shared handle resolves one instance per scope key. - expect(details).toBe(conv) - }) - - it('sessions are isolated: s2 selection never bleeds into s1', () => { - const b = bench() - - const one = storeFor(b, 'conversation.session', sid('s1')) - const two = storeFor(b, 'conversation.session', sid('s2')) - expect(two).not.toBe(one) - one.actions.select({ turnSeq: 1, callId: 'a' }) - two.actions.select({ turnSeq: 9, callId: 'z' }) - expect(one.store.getSnapshot().selection).toEqual({ turnSeq: 1, callId: 'a' }) - expect(two.store.getSnapshot().selection).toEqual({ turnSeq: 9, callId: 'z' }) - }) - - it('a list-projection update keeps instance identity and the selection value', () => { - const b = bench() - const id = sid('s1') - const projection = createSnapshotStore({ displayTitle: 's1' }) - - const store = storeFor(b, 'conversation.session', id) - store.actions.select({ turnSeq: 3, callId: 'c1' }) - store.actions.setDraft('half-typed') - - projection.set({ displayTitle: 'proj-a' }) - expect(projection.getSnapshot().displayTitle).toBe('proj-a') - - const after = storeFor(b, 'conversation.session', id) - expect(after).toBe(store) - expect(after.store.getSnapshot().selection).toEqual({ turnSeq: 3, callId: 'c1' }) - expect(after.store.getSnapshot().draft).toBe('half-typed') - }) - - it('session death buries the instance and its persisted draft', () => { - const b = bench() - - const doomed = storeFor(b, 'conversation.session', sid('s1')) - doomed.actions.setDraft('to be buried') - doomed.actions.select({ turnSeq: 1 }) - expect(localStorage.getItem('dsh.conversation.chat.s1')).not.toBeNull() - - // SessionsService calls this public slot lifecycle seam when the scope dies. - b.slots.pruneStoreScope(sid('s1')) - - // Persisted residue is gone with the session... - expect(localStorage.getItem('dsh.conversation.chat.s1')).toBeNull() - // ...and a re-created same-id session starts from a FRESH instance. - const reborn = storeFor(b, 'conversation.session', sid('s1')) - expect(reborn).not.toBe(doomed) - expect(reborn.store.getSnapshot()).toEqual({ selection: null, draft: '', view: null }) - }) -}) diff --git a/packages/client/ui-conversation/tests/selection-survival.spec.tsx b/packages/client/ui-conversation/tests/selection-survival.spec.tsx new file mode 100644 index 0000000000..70162105a9 --- /dev/null +++ b/packages/client/ui-conversation/tests/selection-survival.spec.tsx @@ -0,0 +1,109 @@ +// @vitest-environment jsdom +/** + * Exercises selection persistence through the real SlotsService store axis; + * component stubs cannot prove per-session identity or disposal. + */ +import { beforeEach, describe, expect, it } from 'vitest' +import type { SessionId } from '@deepseek-ai/dsh-client-runtime/client' +import { SlotTestRuntime } from '@deepseek-ai/dsh-client-test-runtime' +import { createChatStore } from '../src/client/stores.ts' + +const sid = (s: string): SessionId => s as SessionId + +type ChatInstance = ReturnType['create']> + +async function bench() { + const runtime = await SlotTestRuntime.create() + const chat = createChatStore() + // The apply.ts shape: one shared handle across both strict-session slot + // registrations ('conversation.session'/'details'); the session-maybe + // 'conversation' shell carries no store by design. The slots must first + // exist in the ledger — the test root declares them (the AppFrame role). + await runtime.root.declare({ + 'conversation': { kind: 'single', scope: 'session-maybe' }, + 'conversation.session': { kind: 'single', scope: 'session' }, + 'details': { kind: 'single', scope: 'session' }, + }, (_p: { renderSlot?: unknown }) => null) + runtime.slots.register({ name: 'conversation.session', store: chat }, () => null) + runtime.slots.register({ name: 'details', store: chat }, () => null) + runtime.renderRoot() // materializes the host face storeOf resolves through + return { runtime, chat } +} + +/** Resolve the store instance the renderer would hand a slot's component for a session. */ +function storeFor(b: Awaited>, slot: 'conversation.session' | 'details', sessionId: SessionId) { + return b.runtime.storeOf(slot, sessionId) as ChatInstance +} + +beforeEach(() => { + localStorage.clear() +}) + +describe('selection survives on the store seat', () => { + it('one session, two slots: conversation writes, details reads the SAME instance', async () => { + const b = await bench() + + const conv = storeFor(b, 'conversation.session', sid('s1')) + const details = storeFor(b, 'details', sid('s1')) + conv.actions.select({ turnSeq: 3, callId: 'c1' }) + expect(details.store.getSnapshot().selection).toEqual({ turnSeq: 3, callId: 'c1' }) + // Identity, not just value: the shared handle resolves one instance per scope key. + expect(details).toBe(conv) + await b.runtime.dispose() + }) + + it('sessions are isolated: s2 selection never bleeds into s1', async () => { + const b = await bench() + + const one = storeFor(b, 'conversation.session', sid('s1')) + const two = storeFor(b, 'conversation.session', sid('s2')) + expect(two).not.toBe(one) + one.actions.select({ turnSeq: 1, callId: 'a' }) + two.actions.select({ turnSeq: 9, callId: 'z' }) + expect(one.store.getSnapshot().selection).toEqual({ turnSeq: 1, callId: 'a' }) + expect(two.store.getSnapshot().selection).toEqual({ turnSeq: 9, callId: 'z' }) + await b.runtime.dispose() + }) + + it('a list-projection update keeps instance identity and the selection value', async () => { + const b = await bench() + const id = sid('s1') + + const store = storeFor(b, 'conversation.session', id) + store.actions.select({ turnSeq: 3, callId: 'c1' }) + store.actions.setDraft('half-typed') + + // A projection churn elsewhere (list rows re-projected) must not touch + // store identity: drive the runtime's own list observable. + await b.runtime.sessions.add({ id, summary: { displayTitle: 'proj-a' } }) + expect(b.runtime.sessions.list.getSnapshot().byId[id]?.displayTitle).toBe('proj-a') + + const after = storeFor(b, 'conversation.session', id) + expect(after).toBe(store) + expect(after.store.getSnapshot().selection).toEqual({ turnSeq: 3, callId: 'c1' }) + expect(after.store.getSnapshot().draft).toBe('half-typed') + await b.runtime.dispose() + }) + + it('session death buries the instance and its persisted draft', async () => { + const b = await bench() + await b.runtime.sessions.add({ id: 's1' }) + + const doomed = storeFor(b, 'conversation.session', sid('s1')) + doomed.actions.setDraft('to be buried') + doomed.actions.select({ turnSeq: 1 }) + expect(localStorage.getItem('dsh.conversation.chat.s1')).not.toBeNull() + + // TestSessions.remove drives the same public slot lifecycle seam the + // production SessionsService calls when the scope dies (pruneStoreScope). + await b.runtime.sessions.remove('s1') + + // Persisted residue is gone with the session... + expect(localStorage.getItem('dsh.conversation.chat.s1')).toBeNull() + // ...and a re-created same-id session starts from a FRESH instance. + const reborn = storeFor(b, 'conversation.session', sid('s1')) + expect(reborn).not.toBe(doomed) + expect(reborn.store.getSnapshot()).toEqual({ selection: null, draft: '', view: null }) + await b.runtime.dispose() + }) +}) diff --git a/packages/client/ui-conversation/tests/service-orchestration.spec.ts b/packages/client/ui-conversation/tests/service-orchestration.spec.ts index b364dde92a..4897fe5769 100644 --- a/packages/client/ui-conversation/tests/service-orchestration.spec.ts +++ b/packages/client/ui-conversation/tests/service-orchestration.spec.ts @@ -1,39 +1,32 @@ // @vitest-environment jsdom +// ConversationService scope addressing over the runtime's real scope tag: +// TestSessions mints tagged scopes through the production createScope, so the +// service's scopeOf/binding path runs against production resolution (no local +// tag probe). import { Context } from 'cordis' import { describe, expect, it, vi } from 'vitest' -import { scopeOf } from '@deepseek-ai/dsh-client-runtime/client' -import type { SessionId, SessionsService } from '@deepseek-ai/dsh-client-runtime/client' +import { SlotTestRuntime } from '@deepseek-ai/dsh-client-test-runtime' import { ConversationService } from '@deepseek-ai/dsh-client-ui-conversation/client' +import { InputHub } from '../src/client/input/hub.ts' -const sid = (id: string) => id as SessionId -const SCOPE_TAG: symbol = (() => { - const reads: (string | symbol)[] = [] - const proxy = new Proxy(new Context(), { - get(target, property, receiver): unknown { - reads.push(property) - return Reflect.get(target, property, receiver) - }, - }) - void scopeOf(proxy) - return reads.find((value): value is symbol => typeof value === 'symbol')! -})() - -async function bench(withSessions = true) { - const ctx = new Context() +async function bench() { + const runtime = await SlotTestRuntime.create() const prompt = vi.fn(() => Promise.resolve({ ok: true as const, value: { accepted: true as const } })) const cancel = vi.fn(() => Promise.resolve({ ok: true as const, value: { accepted: true as const } })) const loadOlder = vi.fn(() => Promise.resolve()) - const sessions = { - binding: (sessionId: SessionId) => ({ - sessionId, session: { prompt, cancel, loadOlder }, - }), - scopeOf, - } as unknown as SessionsService - if (withSessions) ctx.provide('sessions', sessions) - await ctx.plugin(ConversationService).await() - const root = ctx.get('conversation') as ConversationService - const scoped = ctx.plugin(() => {}).ctx.extend({ [SCOPE_TAG]: sid('s1') }).get('conversation') as ConversationService - return { root, scoped, prompt, cancel, loadOlder } + await runtime.sessions.add({ + id: 's1', + session: { prompt, cancel, loadOlder }, + }) + // config.input is required (the apply shares its hub with the inject + // factories); the bench passes its own instance explicitly. + const fiber = runtime.ctx.plugin(ConversationService, { + input: new InputHub(runtime.ctx), + }) + await fiber.await() + const root = runtime.ctx.get('conversation') as ConversationService + const scoped = runtime.sessions.scope('s1')!.get('conversation') as ConversationService + return { runtime, root, scoped, prompt, cancel, loadOlder } } describe('ConversationService', () => { @@ -45,6 +38,7 @@ describe('ConversationService', () => { expect(b.prompt).toHaveBeenCalledWith([{ type: 'text', text: 'hello' }], 'steer') expect(b.cancel).toHaveBeenCalledOnce() expect(b.loadOlder).toHaveBeenCalledOnce() + await b.runtime.dispose() }) it('folds Session business failures into callback rejections', async () => { @@ -53,12 +47,21 @@ describe('ConversationService', () => { await expect(b.scoped.send('x', 'queue')).rejects.toThrow('conversation.send failed: agent-busy: busy') b.cancel.mockResolvedValueOnce({ ok: false, error: { code: 'internal', message: 'nope', details: {} } } as never) await expect(b.scoped.cancel()).rejects.toThrow('conversation.cancel failed: internal: nope') + await b.runtime.dispose() }) - it('fails loudly from the root scope or without SessionsService', async () => { + it('fails loudly from the root scope, on an unbound session, or without SessionsService', async () => { const b = await bench() await expect(b.root.send('x', 'queue')).rejects.toThrow(/requires a session scope/) - const missing = await bench(false) - await expect(missing.root.send('x', 'queue')).rejects.toThrow(/sessions service unavailable/) + await b.runtime.sessions.remove('s1') + await expect(b.scoped.send('x', 'queue')).rejects.toThrow(/resolved no binding/) + await b.runtime.dispose() + // No SessionsService at all: a bare context (the runtime always provides one). + const bare = new Context() + await bare.plugin(ConversationService, { + input: new InputHub(bare), + }).await() + const orphan = bare.get('conversation') as ConversationService + await expect(orphan.send('x', 'queue')).rejects.toThrow(/sessions service unavailable/) }) }) diff --git a/packages/client/ui-conversation/tsconfig.json b/packages/client/ui-conversation/tsconfig.json index 32ba48e8fe..c08f4b28f5 100644 --- a/packages/client/ui-conversation/tsconfig.json +++ b/packages/client/ui-conversation/tsconfig.json @@ -26,6 +26,9 @@ { "path": "../../session-projection/session-projection" }, + { + "path": "../../plan/plan-mode" + }, { "path": "../../todo/tool-todo" }, diff --git a/packages/client/ui-goal/README.i18n.yaml b/packages/client/ui-goal/README.i18n.yaml new file mode 100644 index 0000000000..9cda25e2e2 --- /dev/null +++ b/packages/client/ui-goal/README.i18n.yaml @@ -0,0 +1,6 @@ +# Bilingual-pair consistency record (docs/i18n/README.md): the git blob hash of each +# side as of the last confirmed-consistent state. Both languages carry equal authority; +# after editing either side, bring the other along and re-record with: +# pnpm run verify-translation-pairing --write packages/client/ui-goal/README.md +README.md: 476096a43532a0bf514cd191585872ef17f65c50 +README.zh.md: 27bd9a2e735cb4895d30eaf3b08dd939a00436fc diff --git a/packages/client/ui-goal/README.md b/packages/client/ui-goal/README.md new file mode 100644 index 0000000000..476096a435 --- /dev/null +++ b/packages/client/ui-goal/README.md @@ -0,0 +1,20 @@ +# @deepseek-ai/dsh-client-ui-goal + +English | [中文](README.zh.md) + +Goal surface plugin, browser half: the `GoalBar` strip in the `conversation.input.dock` list (order 1, tucked against the composer). The live goal arrives through `useProjection('goal')` — the host-computed whole value seeded by the history tail page and updated by `session/projection` frames — so the plugin owns no store, no refresh chain, and no event listener. The slot inject face carries only the three mutation verbs (edit / resume / clear over the `goal.*` wire domain); each reads the CAS ref from the session's current projected value at call time and surfaces the settled RPC error inline (the RPC's compare-and-set is the staleness guard — there is no client fence). Goal creation stays on the `/goal` host command; loading, absent, and completed goals render nothing. + +The `/client` export surface is the plugin body (`apply`/`inject`), the `GoalBar`/`GoalDock` components, and the injected verb face types. + +## Model Experience + +Indirectly, through the `goal.edit`/`goal.resume`/`goal.clear` RPCs the strip's verbs submit: each accepted mutation appends a model-visible `goal/change` context message to the session (the same durable event the projection folds), so the model sees the updated goal state on its next turn. The strip itself adds no prompt content. + +#### KV Cache effect + +None beyond the goal mutation's own context event, which appends to the log tail like any other message. + +## Known Limitations and Deferred Work + +- **Durable phase only** — the projection value deliberately omits process-local activation (armed/disarmed), so the strip cannot distinguish an active-but-disarmed goal from an armed one; resume re-arms through the RPC side. A host-live-value channel is deferred until a real consumer needs it. +- **No keyless snapshot yet** — the assembled-application transcript (boot → projection → GoalBar) is deferred to the post-review cleanup pass recorded on the landing PR. diff --git a/packages/client/ui-goal/README.zh.md b/packages/client/ui-goal/README.zh.md new file mode 100644 index 0000000000..27bd9a2e73 --- /dev/null +++ b/packages/client/ui-goal/README.zh.md @@ -0,0 +1,20 @@ +# @deepseek-ai/dsh-client-ui-goal + +[English](README.md) | 中文 + +Goal 表面插件（浏览器半件）：`conversation.input.dock` 列表中的 `GoalBar` 条带（order 1，紧贴 composer）。活值经 `useProjection('goal')` 到达——host 计算的全量值由历史尾页播种、由 `session/projection` 帧更新——因此本插件不持有 store、不设刷新链、不挂事件监听。slot 注入面只携带三个变更动词（edit / resume / clear，走 `goal.*` 协议域）；每个动词在调用时从会话当前投影值读取 CAS ref，并把结算后的 RPC 错误内联呈现（RPC 的 compare-and-set 即陈旧性防护——客户端没有任何栅栏）。goal 的创建仍归 `/goal` host 命令；加载中、无 goal、已完成三种状态一律不渲染。 + +`/client` 出口面为插件本体（`apply`/`inject`）、`GoalBar`/`GoalDock` 组件与注入动词面类型。 + +## Model Experience + +间接影响：条带动词提交的 `goal.edit`/`goal.resume`/`goal.clear` RPC 每次被接受后，会向会话追加一条模型可见的 `goal/change` 上下文消息（与投影折叠的正是同一条持久事件），模型在下一轮即可看到更新后的 goal 状态。条带自身不添加任何提示词内容。 + +#### KV Cache effect + +除 goal 变更自身的上下文事件（如同任何消息一样追加在日志尾部）外无额外影响。 + +## Known Limitations and Deferred Work + +- **只反映持久 phase** —— 投影值有意省略进程本地的 activation（armed/disarmed），条带无法区分 active-but-disarmed 与 armed 状态；resume 经 RPC 侧重新武装。host 活值通道待出现真实消费方后再议。 +- **暂缺 keyless 快照** —— 组装应用级 transcript（boot → 投影 → GoalBar）推迟到落地 PR 记录的评审后收口批次。 diff --git a/packages/client/ui-goal/package.json b/packages/client/ui-goal/package.json new file mode 100644 index 0000000000..d734539d73 --- /dev/null +++ b/packages/client/ui-goal/package.json @@ -0,0 +1,70 @@ +{ + "name": "@deepseek-ai/dsh-client-ui-goal", + "description": "Session goal surface: GoalBar docked above the composer, read from the goal session projection", + "version": "0.0.1", + "private": true, + "type": "module", + "main": "lib/index.js", + "types": "lib/types/index.d.ts", + "exports": { + ".": { + "types": "./lib/types/index.d.ts", + "default": "./lib/index.js" + }, + "./invariant": { + "types": "./lib/types/invariant.d.ts", + "default": "./lib/invariant.js" + }, + "./client": { + "types": "./lib/types/client/index.d.ts", + "default": "./lib/client.js" + }, + "./src/*": "./src/*", + "./package.json": "./package.json" + }, + "dshClient": { + "inject": [ + "@deepseek-ai/dsh-client-runtime", + "@deepseek-ai/dsh-client-ui-conversation" + ], + "platform": "web" + }, + "scripts": { + "bundle": "tsdown", + "watch": "tsdown --watch" + }, + "license": "BSD-3-Clause", + "peerDependencies": { + "@deepseek-ai/dsh-client-connection": "^0.0.1", + "@deepseek-ai/dsh-client-runtime": "^0.0.1", + "@deepseek-ai/dsh-client-ui-conversation": "^0.0.1", + "@deepseek-ai/dsh-client-ui-primitives": "^0.0.1", + "@deepseek-ai/dsh-client-ui-slots": "^0.0.1", + "@deepseek-ai/dsh-goal": "^0.0.1", + "@deepseek-ai/dsh-invariants": "^0.0.1", + "cordis": "^4.0.0-rc.7", + "react": "^18.2.0" + }, + "devDependencies": { + "@deepseek-ai/dsh-client-connection": "workspace:^", + "@deepseek-ai/dsh-client-runtime": "workspace:^", + "@deepseek-ai/dsh-client-ui-conversation": "workspace:^", + "@deepseek-ai/dsh-client-ui-primitives": "workspace:^", + "@deepseek-ai/dsh-client-ui-slots": "workspace:^", + "@deepseek-ai/dsh-goal": "workspace:^", + "@deepseek-ai/dsh-invariants": "workspace:^", + "@testing-library/react": "^16.1.0", + "@types/react": "~18.3.1", + "cordis": "^4.0.0-rc.7", + "react": "^18.2.0", + "react-dom": "^18.2.0" + }, + "files": [ + "lib/index.js", + "lib/invariant.js", + "lib/client.js", + "lib/types/**/*.d.ts", + "lib/types/**/*.d.ts.map", + "src" + ] +} diff --git a/packages/client/ui-goal/src/client/GoalBar.module.css b/packages/client/ui-goal/src/client/GoalBar.module.css new file mode 100644 index 0000000000..fe07bace1b --- /dev/null +++ b/packages/client/ui-goal/src/client/GoalBar.module.css @@ -0,0 +1,120 @@ +/* GoalBar: the goal strip docked above the composer card. The dock mirrors + InputBar's horizontal geometry (32px side padding, 776px centered cap) + plus the mock's 12px inset, so the bar's edges land 12px inside the + composer card's edges in both the capped and the squeezed regimes. The + negative bottom margin eats InputBar's 8px top padding and tucks the + bar's square bottom edge 2px under the composer card's top edge (the + card, later in DOM order, paints over it). All states share one fixed + 38px height so switching between them never resizes the strip. */ + +.dock { + padding: 0 44px; +} + +.bar { + display: flex; + align-items: center; + gap: 6px; + box-sizing: border-box; + max-width: 752px; + height: 38px; + margin: 0 auto -10px; + padding: 0 14px; + border-radius: 14px 14px 0 0; + /* Translucent hover gray doubles as the mock's #F5F6F7 over the white + base and lifts the strip off the composer card in dark mode. */ + background: var(--dsw-alias-interactive-bg-hover); +} + +.sparkle { + display: inline-flex; + flex: none; + color: var(--dsw-alias-label-tertiary); +} + +.label { + flex: none; + font-size: 13px; + line-height: 20px; + font-weight: 600; + color: var(--dsw-alias-label-primary); +} + +.objective { + flex: 1; + min-width: 0; + overflow: hidden; + font-size: 13px; + line-height: 20px; + color: var(--dsw-alias-label-secondary); + text-overflow: ellipsis; + white-space: nowrap; +} + +.error { + flex: 1; + min-width: 0; + overflow: hidden; + color: var(--dsw-alias-state-error-primary); + font-size: 12px; + line-height: 20px; + text-overflow: ellipsis; + white-space: nowrap; +} + +/* ---- Inline edit form ---- */ + +.objectiveInput { + flex: 1; + min-width: 0; + height: 26px; + padding: 0 8px; + border: 1px solid var(--dsw-alias-border-l2); + border-radius: 6px; + background: var(--dsw-alias-bg-base); + font-size: 13px; + line-height: 20px; + color: var(--dsw-alias-label-primary); + outline: none; +} + +.objectiveInput:focus { + border-color: var(--dsw-alias-state-business-primary); +} + +.objectiveInput::placeholder { + color: var(--dsw-alias-label-caption); +} + +/* ---- Icon actions ---- */ + +.actions { + display: flex; + align-items: center; + gap: 2px; + flex: none; +} + +.iconBtn { + display: inline-flex; + align-items: center; + justify-content: center; + width: 24px; + height: 24px; + padding: 0; + border: none; + border-radius: 6px; + background: transparent; + color: var(--dsw-alias-label-tertiary); + cursor: pointer; +} + +.iconBtn:hover { + background: var(--dsw-alias-interactive-bg-hover); + color: var(--dsw-alias-label-secondary); +} + +.iconBtn:disabled { + opacity: 0.4; + cursor: default; +} diff --git a/packages/client/ui-goal/src/client/GoalBar.tsx b/packages/client/ui-goal/src/client/GoalBar.tsx new file mode 100644 index 0000000000..76308734fc --- /dev/null +++ b/packages/client/ui-goal/src/client/GoalBar.tsx @@ -0,0 +1,161 @@ +/** + * GoalBar: the goal indicator docked above the message composer (input dock + * strip). A present goal shows a sparkle, a phase label, the truncated + * objective, and icon actions — resume when paused, edit (inline form in the + * same strip), and clear. Goal creation lives on the `/goal` command, not + * here: loading (undefined), no goal (null), and complete goals render + * nothing. Live state arrives as the projected whole snapshot; the verbs are + * the injected face. + */ + +import { useCallback, useEffect, useState } from 'react' +import type { GoalSnapshot } from '@deepseek-ai/dsh-goal/client' +import { + IconCheckOutline16, IconCloseOutline16, IconEditOutline16, IconPlayOutline16, IconSparkle16, IconTrashOutline16, +} from '@deepseek-ai/dsh-client-ui-primitives' +import type { GoalActionResult, GoalBarActions } from './slots.ts' +import css from './GoalBar.module.css' + +export interface GoalBarProps extends GoalBarActions { + /** Current goal snapshot; undefined = capability absent or loading, null = no goal set. */ + goal: GoalSnapshot | null | undefined +} + +/** Strip labels per visible phase; complete goals render nothing. */ +const PHASE_LABELS = { + active: 'Ongoing Goal', + paused: 'Paused Goal', + blocked: 'Blocked Goal', +} as const + +export function GoalBar({ goal, onEdit, onResume, onClear }: GoalBarProps) { + const [editing, setEditing] = useState(false) + const [draft, setDraft] = useState('') + const [pending, setPending] = useState(false) + const [actionError, setActionError] = useState(null) + + // A new goal identity (cleared/completed/replaced externally) invalidates the local edit + // state: without the reset a surviving draft's Enter would write over the NEW goal. + const goalId = goal?.id + useEffect(() => { + setEditing(false) + setActionError(null) + }, [goalId]) + + const handleEdit = useCallback(async () => { + const trimmed = draft.trim() + if (trimmed === '') return + setPending(true) + setActionError(null) + const result = await onEdit(trimmed) + setPending(false) + if (result.ok) { + setEditing(false) + } else { + setActionError(`${result.error.message} (${result.error.code})`) + } + }, [draft, onEdit]) + + const runAction = useCallback(async (action: () => Promise) => { + setPending(true) + setActionError(null) + const result = await action() + setPending(false) + if (!result.ok) setActionError(`${result.error.message} (${result.error.code})`) + }, []) + + // Loading, absent, and complete goals have no strip at all. + if (goal === undefined || goal === null || goal.phase === 'complete') return null + + if (editing) { + return ( +
+
+ { setDraft(e.target.value) }} + onKeyDown={(e) => { + if (e.key === 'Enter') void handleEdit() + if (e.key === 'Escape') setEditing(false) + }} + autoFocus + /> + {actionError !== null && {actionError}} +
+ + +
+
+
+ ) + } + + const title = goal.phase === 'blocked' ? goal.blockedReason?.message : undefined + return ( +
+
+ + {PHASE_LABELS[goal.phase]} + {goal.objective} + {actionError !== null && {actionError}} +
+ {goal.phase === 'paused' && ( + + )} + + +
+
+
+ ) +} + +/** Full props of the dock entry: InputZone owner share + session standard kit + injected verbs. */ +export type GoalDockProps = import('@deepseek-ai/dsh-client-ui-slots').PropsRuntime<'conversation.input.dock'> & GoalBarActions + +/** Dock adapter: reads the host-computed 'goal' projection (whole value; absent or null renders nothing). */ +export function GoalDock({ useProjection, onEdit, onResume, onClear }: GoalDockProps) { + const projection = useProjection('goal') + return ( + + ) +} diff --git a/packages/client/ui-goal/src/client/index.ts b/packages/client/ui-goal/src/client/index.ts new file mode 100644 index 0000000000..9c9d3f298d --- /dev/null +++ b/packages/client/ui-goal/src/client/index.ts @@ -0,0 +1,82 @@ +/** + * Goal surface plugin, browser half: the GoalBar entry in the + * conversation.input.dock strip. Projection-mode surface — the live goal + * arrives through `useProjection('goal')` (seeded by the history tail page, + * updated by session/projection frames), so this plugin owns no store, no + * refresh chain, and no event listener. The inject face carries only the + * three mutation verbs (edit/resume/clear over the goal.* wire domain); + * their CAS ref reads the session's current projected value at call time. + * Goal creation stays on the /goal host command. + */ +import type { ConnectionHandle, GoalRef, SessionId } from '@deepseek-ai/dsh-client-connection/client' +import type { RpcResult } from '@deepseek-ai/dsh-client-connection/client' +import type { ClientContext } from '@deepseek-ai/dsh-client-runtime/client' +// Type-only: pulls the ui-conversation SlotMap merge (the input.dock entry). +import type {} from '@deepseek-ai/dsh-client-ui-conversation/client' +// Type-only: the `goal` SessionProjectionMap key merge (single source, the domain's pure outlet). +import type { GoalProjection } from '@deepseek-ai/dsh-goal/client' +import type { GoalActionResult, GoalBarActions } from './slots.ts' +import { GoalDock } from './GoalBar.tsx' + +export { GoalBar, GoalDock } from './GoalBar.tsx' +export type { GoalActionResult, GoalBarActions } from './slots.ts' + +/** Required services: slots for the dock entry, sessions for the projected ref, connection for the wire verbs. */ +export const inject = ['slots', 'sessions', 'connection'] + +/** Map one settled RPC result onto the strip's inline-render shape. */ +function settle(result: RpcResult): GoalActionResult { + if (result.ok) return { ok: true } + return { ok: false, error: { code: result.error.code, message: result.error.message } } +} + +/** + * Client plugin body: the GoalBar dock entry with its mutation verbs. + * @param ctx - client root context. + */ +export function apply(ctx: ClientContext): void { + const { goals } = (ctx.get('connection') as ConnectionHandle).api + + // Conditional mount: 'conversation.input.dock' is declared by the + // conversation entry; the conversation service being up is the + // registration-safe signal (the TodoDock/QueueDock seam). + ctx.inject(['slots', 'conversation', 'sessions'], (scope: ClientContext) => { + const sessions = scope.sessions + + /** The session's current projected CAS ref, read at verb call time (no staleness fence: the RPC's CAS is the guard). */ + const refOf = (sessionId: SessionId): GoalRef | undefined => { + const face = sessions.binding(sessionId)?.session.projections.faceOf('goal') + const projection = face?.getSnapshot() as GoalProjection | null | undefined + if (projection == null) return undefined + return { id: projection.goal.id, revision: projection.goal.revision } + } + + const noCurrentGoal: GoalActionResult = { + ok: false, + error: { code: 'no-current-goal', message: 'no current goal to mutate' }, + } + + scope.effect(() => scope.slots.register({ + name: 'conversation.input.dock', + id: 'goal', + order: 1, + inject: (sessionId): GoalBarActions => ({ + onEdit: async (objective) => { + const ref = refOf(sessionId) + if (ref === undefined) return noCurrentGoal + return settle((await goals.edit({ sessionId, ref, objective })).result) + }, + onResume: async () => { + const ref = refOf(sessionId) + if (ref === undefined) return noCurrentGoal + return settle((await goals.resume({ sessionId, ref })).result) + }, + onClear: async () => { + const ref = refOf(sessionId) + if (ref === undefined) return noCurrentGoal + return settle((await goals.clear({ sessionId, ref })).result) + }, + }), + }, GoalDock), 'ui-goal: GoalBar dock registration') + }) +} diff --git a/packages/client/ui-goal/src/client/slots.ts b/packages/client/ui-goal/src/client/slots.ts new file mode 100644 index 0000000000..5791f26373 --- /dev/null +++ b/packages/client/ui-goal/src/client/slots.ts @@ -0,0 +1,26 @@ +/** + * GoalBar's injected face. The target 'conversation.input.dock' slot is + * declared (children table) and typed by ui-conversation; this package only + * contributes the entry, so no SlotMap merge lives here. The live goal value + * is NOT part of this face — it arrives through `useProjection('goal')` + * (the framework standard kit); inject carries only the mutation verbs + * (callbacks from inject, live state from useProjection). + */ + +/** Settled outcome of one goal mutation, rendered inline by the strip. */ +export type GoalActionResult = + | { ok: true } + | { ok: false; error: { code: string; message: string } } + +/** Injected business face of the GoalBar dock entry: the mutation verbs (function properties: the strip destructures them freely). */ +export interface GoalBarActions { + /** + * Replace the current goal's objective (CAS on the projected ref). + * @param objective - replacement objective text. + */ + onEdit: (objective: string) => Promise + /** Resume a paused goal. */ + onResume: () => Promise + /** Clear the current goal (tombstone). */ + onClear: () => Promise +} diff --git a/packages/client/ui-goal/src/css-modules.d.ts b/packages/client/ui-goal/src/css-modules.d.ts new file mode 100644 index 0000000000..bc5e482353 --- /dev/null +++ b/packages/client/ui-goal/src/css-modules.d.ts @@ -0,0 +1,6 @@ +declare module '*.module.css' { + const classes: Record + export default classes +} + +declare module '*.css' diff --git a/packages/client/ui-goal/src/index.ts b/packages/client/ui-goal/src/index.ts new file mode 100644 index 0000000000..780cea398e --- /dev/null +++ b/packages/client/ui-goal/src/index.ts @@ -0,0 +1,9 @@ +/** + * Goal surface plugin, node half. Pure UI plugin: the empty apply exists so + * the plugin appears in the host cordis.yml / Loader; the browser half + * ships via exports["./client"], discovered through the package.json + * dshClient declaration. + */ + +/** Host plugin body — no host-side behavior for this surface plugin. */ +export function apply(): void {} diff --git a/packages/client/ui-goal/src/invariant.ts b/packages/client/ui-goal/src/invariant.ts new file mode 100644 index 0000000000..2120600664 --- /dev/null +++ b/packages/client/ui-goal/src/invariant.ts @@ -0,0 +1,32 @@ +/** + * Package-owned invariant companion for `@deepseek-ai/dsh-client-ui-goal`. + * @module @deepseek-ai/dsh-client-ui-goal/invariant + */ + +/* jscpd:ignore-start */ +import type { Context } from 'cordis' +import type { InvariantInstaller } from '@deepseek-ai/dsh-invariants' + +const PACKAGE_NAME = '@deepseek-ai/dsh-client-ui-goal' + +/** Cordis companion plugin name. */ +export const name = 'client-ui-goal-invariant' +/** Service required before the companion can reserve package ownership. */ +export const inject = ['invariants'] + +/** + * No runtime invariant: a single GoalBar dock registration whose disposal is + * proven by the HMR-safety spec — the plugin owns no store (state arrives on + * the goal projection), emits no cordis events, and holds no cross-plugin + * mutable state. + */ +const install: InvariantInstaller = () => {} + +/** + * Register this package's invariant companion. + * @param ctx - Cordis context carrying the invariant service. + * @returns the installed registration's disposer after setup succeeds. + */ +export const apply = (ctx: Context): Promise<() => void> => + Promise.resolve(ctx.invariants.register(PACKAGE_NAME, install)) +/* jscpd:ignore-end */ diff --git a/packages/client/ui-goal/tests/browser-plugin.spec.tsx b/packages/client/ui-goal/tests/browser-plugin.spec.tsx new file mode 100644 index 0000000000..e1eae47529 --- /dev/null +++ b/packages/client/ui-goal/tests/browser-plugin.spec.tsx @@ -0,0 +1,170 @@ +// @vitest-environment jsdom +/** + * ui-goal browser half on a real cordis Context with fake slots/connection/ + * sessions faces: the plugin registers the GoalBar dock entry at + * conversation.input.dock, the inject face's three verbs read the CAS ref + * from the session's CURRENT projected value at call time (no fence — the + * RPC's compare-and-set is the guard), a missing projection short-circuits + * to the no-current-goal error without touching the wire, and RPC errors + * map onto the inline-render result shape. Registration disposal rides the + * plugin fiber (HMR safety). The node half and the invariant companion are + * exercised over the same Context. + */ +import { Context } from 'cordis' +import { describe, expect, it, vi } from 'vitest' +import { cleanup, render } from '@testing-library/react' +import { afterEach } from 'vitest' +import type { SessionId } from '@deepseek-ai/dsh-client-runtime/client' +import type { GoalProjection } from '@deepseek-ai/dsh-goal/client' +import type { GoalBarActions } from '../src/client/slots.ts' +import { apply, inject } from '../src/client/index.ts' +import { GoalDock } from '../src/client/GoalBar.tsx' +import { apply as nodeApply } from '../src/index.ts' + +afterEach(cleanup) + +const sid = (k: string): SessionId => k as SessionId + +function makeProjection(revision = 3): GoalProjection { + return { + goal: { + id: 'g-1' as GoalProjection['goal']['id'], + revision, + objective: 'Ship it', + phase: 'active', + maxGoalRounds: 8, + }, + roundsStarted: 1, + createdAt: 10, + updatedAt: 20, + } +} + +/** Boot the plugin over fake faces; goals verbs record payloads and answer per the script. */ +function bench(options: { projection?: GoalProjection | null | undefined; failWith?: { code: string; message: string } } = {}) { + const ctx = new Context() + const calls: { method: string; payload: unknown }[] = [] + function answer(method: string, value: T) { + return (payload: unknown) => { + calls.push({ method, payload }) + return Promise.resolve({ + result: options.failWith === undefined + ? { ok: true as const, value } + : { ok: false as const, error: { ...options.failWith, details: {} } }, + }) + } + } + const ref = { id: 'g-1', revision: 3 } + ctx.provide('connection', { api: { goals: { + edit: answer('goal.edit', { ref }), + resume: answer('goal.resume', { ref }), + clear: answer('goal.clear', { cleared: true as const }), + } } }) + const entries = new Map GoalBarActions }>() + ctx.provide('slots', { + register(reg: { name: string; id?: string; order?: number; inject?: (sessionId: SessionId) => GoalBarActions }) { + entries.set(reg.name, reg) + return () => { entries.delete(reg.name) } + }, + }) + ctx.provide('conversation', {}) + ctx.provide('sessions', { + binding: (id: SessionId) => ({ + sessionId: id, + session: { projections: { faceOf: (key: string) => ({ + getSnapshot: () => (key === 'goal' ? options.projection : undefined), + subscribe: () => () => {}, + }) } }, + ctx, + }), + }) + const fiber = ctx.plugin({ inject: [...inject], apply }) + return { + ctx, + fiber, + calls, + entry: () => entries.get('conversation.input.dock'), + } +} + +describe('ui-goal browser plugin', () => { + it('registers the GoalBar dock entry with the documented id and order', async () => { + const b = bench() + await b.fiber.await() + expect(b.entry()).toMatchObject({ id: 'goal', order: 1 }) + expect(b.entry()?.inject).toBeTypeOf('function') + }) + + it('verbs read the CAS ref from the current projected value at call time', async () => { + const b = bench({ projection: makeProjection(5) }) + await b.fiber.await() + const verbs = b.entry()!.inject!(sid('s1')) + expect(await verbs.onEdit('New objective')).toEqual({ ok: true }) + expect(await verbs.onResume()).toEqual({ ok: true }) + expect(await verbs.onClear()).toEqual({ ok: true }) + expect(b.calls.map(c => c.method)).toEqual(['goal.edit', 'goal.resume', 'goal.clear']) + const ref = { id: 'g-1', revision: 5 } + expect(b.calls[0]?.payload).toEqual({ sessionId: 's1', ref, objective: 'New objective' }) + expect(b.calls[1]?.payload).toEqual({ sessionId: 's1', ref }) + expect(b.calls[2]?.payload).toEqual({ sessionId: 's1', ref }) + }) + + it('a null or absent projection short-circuits every verb without touching the wire', async () => { + for (const projection of [null, undefined]) { + const b = bench({ projection }) + await b.fiber.await() + const verbs = b.entry()!.inject!(sid('s1')) + for (const result of [await verbs.onEdit('x'), await verbs.onResume(), await verbs.onClear()]) { + expect(result).toEqual({ ok: false, error: { code: 'no-current-goal', message: 'no current goal to mutate' } }) + } + expect(b.calls).toHaveLength(0) + } + }) + + it('maps a settled RPC error onto the inline-render shape', async () => { + const b = bench({ projection: makeProjection(), failWith: { code: 'internal', message: 'stale revision' } }) + await b.fiber.await() + const verbs = b.entry()!.inject!(sid('s1')) + expect(await verbs.onEdit('x')).toEqual({ ok: false, error: { code: 'internal', message: 'stale revision' } }) + }) + + it('drops the dock entry when the plugin fiber unloads (HMR safety)', async () => { + const b = bench() + await b.fiber.await() + expect(b.entry()).toBeDefined() + await b.fiber.dispose() + expect(b.entry()).toBeUndefined() + }) +}) + +describe('GoalDock adapter', () => { + it('renders the projected goal snapshot and nothing for absent/null', () => { + const projection = makeProjection() + const useProjection = vi.fn(() => projection) + const actions: GoalBarActions = { + onEdit: () => Promise.resolve({ ok: true }), + onResume: () => Promise.resolve({ ok: true }), + onClear: () => Promise.resolve({ ok: true }), + } + const dockProps = (up: () => GoalProjection | null | undefined) => + ({ useProjection: up, ...actions }) as unknown as Parameters[0] + const shown = render() + expect(shown.getByText('Ship it')).toBeTruthy() + cleanup() + + const empty = render( null)} />) + expect(empty.container.firstChild).toBeNull() + cleanup() + + const absent = render( undefined)} />) + expect(absent.container.firstChild).toBeNull() + }) +}) + +describe('ui-goal node half', () => { + // The invariant companion is mounted by the vitest-wide invariant host on + // every Context this suite creates; its registration is covered there. + it('the node apply is an inert loader seat', () => { + expect(() => { nodeApply() }).not.toThrow() + }) +}) diff --git a/packages/client/ui-goal/tests/goalbar.spec.tsx b/packages/client/ui-goal/tests/goalbar.spec.tsx new file mode 100644 index 0000000000..ece447f8b0 --- /dev/null +++ b/packages/client/ui-goal/tests/goalbar.spec.tsx @@ -0,0 +1,170 @@ +// @vitest-environment jsdom +// GoalBar behavior: the docked strip above the composer — phase labels, +// inline edit form, and resume/clear icon actions — driven purely through +// props, no wire. Loading, absent, and complete goals render nothing. + +import { cleanup, fireEvent, render, screen, waitFor } from '@testing-library/react' +import { afterEach, describe, expect, it, vi } from 'vitest' +import type { GoalSnapshot } from '@deepseek-ai/dsh-goal/client' +import { GoalBar } from '../src/client/GoalBar.tsx' +import type { GoalBarActions } from '../src/client/slots.ts' + +afterEach(cleanup) + +function makeGoal(over: Partial = {}): GoalSnapshot { + return { + id: 'g1' as GoalSnapshot['id'], + revision: 1, + objective: 'Ship the redesign', + phase: 'active', + maxGoalRounds: 4, + ...over, + } +} + +function makeActions() { + return { + onEdit: vi.fn(() => Promise.resolve({ ok: true })), + onResume: vi.fn(() => Promise.resolve({ ok: true })), + onClear: vi.fn(() => Promise.resolve({ ok: true })), + } satisfies GoalBarActions +} + +describe('GoalBar', () => { + it('renders nothing while loading, absent, or when the goal is complete', () => { + const actions = makeActions() + const loading = render() + expect(loading.container.firstChild).toBeNull() + cleanup() + + const absent = render() + expect(absent.container.firstChild).toBeNull() + cleanup() + + const complete = render() + expect(complete.container.firstChild).toBeNull() + }) + + it('active goal: sparkle, "Ongoing Goal", truncated objective, edit and clear actions', () => { + const actions = makeActions() + render() + expect(screen.getByText('Ongoing Goal')).toBeTruthy() + expect(screen.getByText('Ship the redesign')).toBeTruthy() + fireEvent.click(screen.getByRole('button', { name: 'Clear goal' })) + expect(actions.onClear).toHaveBeenCalledTimes(1) + }) + + it('edit swaps the strip for a prefilled form; Enter saves, empty stays disabled', async () => { + const actions = makeActions() + render() + fireEvent.click(screen.getByRole('button', { name: 'Edit goal' })) + const box = screen.getByRole('textbox', { name: 'Goal objective' }) + expect(box).toHaveProperty('value', 'Ship the redesign') + + fireEvent.change(box, { target: { value: ' ' } }) + expect(screen.getByRole('button', { name: 'Save goal' })).toHaveProperty('disabled', true) + + fireEvent.change(box, { target: { value: 'Ship v2' } }) + fireEvent.keyDown(box, { key: 'Enter' }) + expect(actions.onEdit).toHaveBeenCalledWith('Ship v2') + await waitFor(() => { expect(screen.getByText('Ongoing Goal')).toBeTruthy() }) + }) + + it('Esc cancels the edit without calling onEdit', () => { + const actions = makeActions() + render() + fireEvent.click(screen.getByRole('button', { name: 'Edit goal' })) + fireEvent.keyDown(screen.getByRole('textbox', { name: 'Goal objective' }), { key: 'Escape' }) + expect(actions.onEdit).not.toHaveBeenCalled() + expect(screen.getByText('Ongoing Goal')).toBeTruthy() + }) + + it('the cancel button exits the form and drops the draft (re-edit starts from the objective)', () => { + const actions = makeActions() + render() + fireEvent.click(screen.getByRole('button', { name: 'Edit goal' })) + fireEvent.change(screen.getByRole('textbox', { name: 'Goal objective' }), { target: { value: 'abandoned draft' } }) + fireEvent.click(screen.getByRole('button', { name: 'Cancel edit' })) + expect(actions.onEdit).not.toHaveBeenCalled() + expect(screen.getByText('Ongoing Goal')).toBeTruthy() + + fireEvent.click(screen.getByRole('button', { name: 'Edit goal' })) + expect(screen.getByRole('textbox', { name: 'Goal objective' })).toHaveProperty('value', 'Ship the redesign') + }) + + it('Enter with a blank draft neither saves nor closes the form', () => { + const actions = makeActions() + render() + fireEvent.click(screen.getByRole('button', { name: 'Edit goal' })) + const box = screen.getByRole('textbox', { name: 'Goal objective' }) + fireEvent.change(box, { target: { value: ' ' } }) + fireEvent.keyDown(box, { key: 'Enter' }) + expect(actions.onEdit).not.toHaveBeenCalled() + expect(screen.getByRole('textbox', { name: 'Goal objective' })).toBeTruthy() + }) + + it('paused goal: "Paused Goal" with a resume action before edit', () => { + const actions = makeActions() + render() + expect(screen.getByText('Paused Goal')).toBeTruthy() + fireEvent.click(screen.getByRole('button', { name: 'Resume goal' })) + expect(actions.onResume).toHaveBeenCalledTimes(1) + }) + + it('a new goal identity drops the edit form (no stale draft over the new goal)', () => { + const actions = makeActions() + const { rerender } = render() + fireEvent.click(screen.getByRole('button', { name: 'Edit goal' })) + fireEvent.change(screen.getByRole('textbox', { name: 'Goal objective' }), { target: { value: 'stale draft' } }) + + rerender() + expect(screen.queryByRole('textbox')).toBeNull() + expect(screen.getByText('Ongoing Goal')).toBeTruthy() + expect(screen.getByText('New goal')).toBeTruthy() + + rerender() + expect(screen.queryByText('Ongoing Goal')).toBeNull() + }) + + it('blocked goal: "Blocked Goal" with the block reason as the strip tooltip', () => { + const actions = makeActions() + const goal = makeGoal({ phase: 'blocked', blockedReason: { code: 'stalled', message: 'No progress in 3 rounds' } }) + render() + expect(screen.getByText('Blocked Goal')).toBeTruthy() + expect(screen.getByText('Blocked Goal').closest('[title]')?.getAttribute('title')).toBe('No progress in 3 rounds') + }) + + it('blocked goal without a reason carries no tooltip', () => { + const actions = makeActions() + render() + expect(screen.getByText('Blocked Goal')).toBeTruthy() + expect(screen.getByText('Blocked Goal').closest('[title]')).toBeNull() + }) + + it('keeps the edit draft open and reports a failed save', async () => { + const actions = makeActions() + actions.onEdit.mockResolvedValue({ ok: false, error: { code: 'agent-busy', message: 'stale revision' } }) + render() + fireEvent.click(screen.getByRole('button', { name: 'Edit goal' })) + const box = screen.getByRole('textbox', { name: 'Goal objective' }) + fireEvent.change(box, { target: { value: 'retry this draft' } }) + fireEvent.click(screen.getByRole('button', { name: 'Save goal' })) + + expect((await screen.findByRole('alert')).textContent).toBe('stale revision (agent-busy)') + expect(screen.getByRole('textbox', { name: 'Goal objective' })).toHaveProperty('value', 'retry this draft') + }) + + it('reports resume and clear failures without hiding the goal', async () => { + const actions = makeActions() + actions.onResume.mockResolvedValue({ ok: false, error: { code: 'internal', message: 'resume failed' } }) + const { rerender } = render() + fireEvent.click(screen.getByRole('button', { name: 'Resume goal' })) + expect((await screen.findByRole('alert')).textContent).toBe('resume failed (internal)') + + actions.onClear.mockResolvedValue({ ok: false, error: { code: 'agent-busy', message: 'clear failed' } }) + rerender() + fireEvent.click(screen.getByRole('button', { name: 'Clear goal' })) + expect((await screen.findByRole('alert')).textContent).toBe('clear failed (agent-busy)') + expect(screen.getByText('Ship the redesign')).toBeTruthy() + }) +}) diff --git a/packages/client/ui-goal/tsconfig.json b/packages/client/ui-goal/tsconfig.json new file mode 100644 index 0000000000..d1d008e92b --- /dev/null +++ b/packages/client/ui-goal/tsconfig.json @@ -0,0 +1,36 @@ +{ + "extends": "../../../tsconfig.base.client.json", + "compilerOptions": { + "rootDir": "src", + "outDir": "lib/types" + }, + "include": [ + "src" + ], + "references": [ + { + "path": "../../../vendor/cordis" + }, + { + "path": "../connection" + }, + { + "path": "../runtime" + }, + { + "path": "../ui-conversation" + }, + { + "path": "../ui-primitives" + }, + { + "path": "../ui-slots" + }, + { + "path": "../../goal/goal" + }, + { + "path": "../../support/invariants" + } + ] +} diff --git a/packages/client/ui-goal/tsdown.config.ts b/packages/client/ui-goal/tsdown.config.ts new file mode 100644 index 0000000000..a01975f78a --- /dev/null +++ b/packages/client/ui-goal/tsdown.config.ts @@ -0,0 +1,3 @@ +import { clientBundle } from '../tsdown.client.ts' + +export default clientBundle('@deepseek-ai/dsh-client-ui-goal', ['lib/types/index.js', 'lib/types/invariant.js']) diff --git a/packages/client/ui-layout/src/client/index.ts b/packages/client/ui-layout/src/client/index.ts index 2dd8aafb4e..bb78596a0b 100644 --- a/packages/client/ui-layout/src/client/index.ts +++ b/packages/client/ui-layout/src/client/index.ts @@ -17,14 +17,16 @@ import { ThemePresenter } from './theme-presenter.ts' // Contract surface only (export-convergence rule: cross-package consumers // keep a symbol exported; test-only/package-internal symbols live off /src). -// LayoutService: the ctx.layout service class (consumers type against it). +// ILayout: the ctx.layout face consumers and test fakes type against. // OwnerShare contracts below are the render-side halves registrants compose // against; the frame components and the store factory are package-internal. export { LayoutService } from './service.ts' +export type { ILayout } from './service.ts' declare module 'cordis' { interface Context { - layout: LayoutService + /** The outward face only; the concrete service stays inside this plugin. */ + layout: import('./service.ts').ILayout } } diff --git a/packages/client/ui-layout/src/client/service.ts b/packages/client/ui-layout/src/client/service.ts index 6b80644855..ecfd862a03 100644 --- a/packages/client/ui-layout/src/client/service.ts +++ b/packages/client/ui-layout/src/client/service.ts @@ -14,8 +14,23 @@ import type { createLayoutStore } from './stores.ts' /** The layout store's bound action set (framework-baked, draft params peeled). */ export type PanelActions = BoundActions> +/** + * The outward layout face (`ctx.layout`): the panel transitions other + * plugins may trigger — and exactly what a test fake must supply. The + * attachPanels wiring hook stays on the concrete class (root-entry assembly + * only). + */ +export interface ILayout { + /** Toggle the sidebar panel (closed ⟷ contract default width). */ + toggleSidebar(): void + /** Open the details panel (no-op when already open). */ + openDetails(): void + /** Close the details panel. */ + closeDetails(): void +} + /** Cross-plugin panel-action face (ctx.layout). */ -export class LayoutService { +export class LayoutService implements ILayout { #panels: PanelActions | undefined /** diff --git a/packages/client/ui-plan/README.i18n.yaml b/packages/client/ui-plan/README.i18n.yaml new file mode 100644 index 0000000000..f7572c9bfa --- /dev/null +++ b/packages/client/ui-plan/README.i18n.yaml @@ -0,0 +1,6 @@ +# Bilingual-pair consistency record (docs/i18n/README.md): the git blob hash of each +# side as of the last confirmed-consistent state. Both languages carry equal authority; +# after editing either side, bring the other along and re-record with: +# pnpm run verify-translation-pairing --write packages/client/ui-plan/README.md +README.md: de43ce66d17498d31e05f8c64092ea0843103054 +README.zh.md: b4d2f4fd1a6d45f814d4a20195434f34d207e9c8 diff --git a/packages/client/ui-plan/README.md b/packages/client/ui-plan/README.md new file mode 100644 index 0000000000..de43ce66d1 --- /dev/null +++ b/packages/client/ui-plan/README.md @@ -0,0 +1,25 @@ +# @deepseek-ai/dsh-client-ui-plan + +English | [中文](README.zh.md) + +Plan-mode status chip, a pure browser surface plugin. The browser half occupies the conversation-declared `conversation.input.plan` single seat (to the right of the access-mode control); the node half is an empty apply (the roster row). Plan behavior itself — the `/plan` command, the boundary-or-idle-committed `plan/mode` state, the `plan` projection unit, and the policy section — is owned by [`@deepseek-ai/dsh-plan-mode`](../../plan/plan-mode/README.md), composed independently on the host roster. + +Plan mode is entered through the `/plan` command only; there is no UI control that turns it on. While the host-computed `plan` projection's effective target is plan mode (`pending ? !active : active` — a folded host value, not client optimism, so an arriving frame corrects the chip either way), the seat renders a read-only "Plan" chip whose hover × executes `/plan off` through `command.execute`; otherwise the seat stays empty — a host without plan-mode (or a Draft with no session) shows nothing. While plan mode is the effective target, the composer textarea's placeholder switches to "describe your task to generate plan" (rendered by the composer from the same projection; owner-supplied placeholders win). + +The chip carries the accessible description "Plan mode on, press to turn off". Admission failures (`matched: false`, business errors, transport faults) surface as an inline error and the chip stays until the projection confirms the exit. + +The model exits plan mode through the stable `exit_plan_mode` tool; its plan review uses the composed Web question channel. + +## Model Experience + +Indirectly, through the `/plan off` command line the chip dispatches: `@deepseek-ai/dsh-plan-mode` owns the model-visible policy section, the exit-tool schema, and the logged state that line drives, while this package only renders the projection and sends what a user could equally type. + +#### KV Cache effect + +Entering or leaving plan mode changes the active `plan:policy` system-prompt section and therefore the request prefix; the chip itself adds no prompt content. + +## Known Limitations and Deferred Work + +- **Plan mode is guidance, not an execution sandbox** — deployments that require enforced read-only planning must compose the independent sandbox and approval policies. +- **The chip belongs to the default composer** — a pending whole-composer interaction such as plan review temporarily replaces the InputBar and its chip. +- **No UI entry point** — plan mode is entered by typing `/plan`; a session with the capability but inactive mode shows no affordance in the tool row. diff --git a/packages/client/ui-plan/README.zh.md b/packages/client/ui-plan/README.zh.md new file mode 100644 index 0000000000..b4d2f4fd1a --- /dev/null +++ b/packages/client/ui-plan/README.zh.md @@ -0,0 +1,25 @@ +# @deepseek-ai/dsh-client-ui-plan + +[English](README.md) | 中文 + +Plan mode 状态徽章，纯浏览器 surface 插件。浏览器侧占据会话声明的 `conversation.input.plan` 单座（位于 access 模式控件右侧）；node 侧是空 apply（roster 行）。plan 行为本身——`/plan` 命令、边界或空闲即时提交的 `plan/mode` 状态、`plan` 投影单元与 policy 段——归 [`@deepseek-ai/dsh-plan-mode`](../../plan/plan-mode/README.md) 所有，由 host roster 独立组合。 + +plan mode 只经 `/plan` 命令进入；UI 上没有打开它的控件。当 host 计算的 `plan` 投影有效目标为 plan mode 时（`pending ? !active : active`——折叠的 host 值而非客户端乐观态，帧到达即自动纠正），座位渲染一个只读 "Plan" chip，hover 出现的 × 经 `command.execute` 执行 `/plan off`；否则座位保持为空——未组合 plan-mode 的 host（或尚无会话的 Draft）不显示任何内容。plan mode 为有效目标期间，composer 文本框的 placeholder 切换为 "describe your task to generate plan"（由 composer 从同一投影渲染；owner 提供的 placeholder 优先）。 + +chip 携带无障碍描述 "Plan mode on, press to turn off"。准入失败（`matched: false`、业务错误、传输故障）以内联错误呈现，chip 保持显示直至投影确认退出。 + +模型通过稳定的 `exit_plan_mode` 工具退出 plan mode；其 plan 评审走已组合的 Web question 通道。 + +## 模型体验 + +间接地，通过 chip 派发的 `/plan off` 命令行：`@deepseek-ai/dsh-plan-mode` 拥有该命令行驱动的模型可见 policy 段、退出工具 schema 与已记录状态，本包只渲染投影并发送用户同样可以手敲的内容。 + +#### KV 缓存效应 + +进入或离开 plan mode 会改变活跃的 `plan:policy` 系统提示词段，因此改变请求前缀；chip 本身不添加任何提示词内容。 + +## 已知局限与延后工作 + +- **Plan mode 是引导而非执行沙箱**——需要强制只读规划的部署必须组合独立的沙箱与审批策略。 +- **chip 属于默认编辑器**——待处理的整编辑器交互（如 plan 评审）会临时取代 InputBar 及其 chip。 +- **无 UI 进入点**——plan mode 靠敲 `/plan` 进入；有能力但未激活的会话在工具行不显示任何入口。 diff --git a/packages/client/ui-plan/package.json b/packages/client/ui-plan/package.json new file mode 100644 index 0000000000..a2aec47985 --- /dev/null +++ b/packages/client/ui-plan/package.json @@ -0,0 +1,67 @@ +{ + "name": "@deepseek-ai/dsh-client-ui-plan", + "description": "Plan-mode composer control: the conversation.input.plan seat over the plan projection and the /plan command channel", + "version": "0.0.1", + "private": true, + "type": "module", + "main": "lib/index.js", + "types": "lib/types/index.d.ts", + "exports": { + ".": { + "types": "./lib/types/index.d.ts", + "default": "./lib/index.js" + }, + "./invariant": { + "types": "./lib/types/invariant.d.ts", + "default": "./lib/invariant.js" + }, + "./client": { + "types": "./lib/types/client/index.d.ts", + "default": "./lib/client.js" + }, + "./src/*": "./src/*", + "./package.json": "./package.json" + }, + "dshClient": { + "inject": [ + "@deepseek-ai/dsh-client-connection", + "@deepseek-ai/dsh-client-ui-conversation" + ], + "platform": "web" + }, + "scripts": { + "bundle": "tsdown", + "watch": "tsdown --watch" + }, + "license": "BSD-3-Clause", + "peerDependencies": { + "@deepseek-ai/dsh-client-connection": "^0.0.1", + "@deepseek-ai/dsh-client-runtime": "^0.0.1", + "@deepseek-ai/dsh-client-ui-conversation": "^0.0.1", + "@deepseek-ai/dsh-client-ui-slots": "^0.0.1", + "@deepseek-ai/dsh-invariants": "^0.0.1", + "@deepseek-ai/dsh-plan-mode": "^0.0.1", + "cordis": "^4.0.0-rc.7", + "react": "^18.2.0" + }, + "devDependencies": { + "@deepseek-ai/dsh-client-connection": "workspace:^", + "@deepseek-ai/dsh-client-runtime": "workspace:^", + "@deepseek-ai/dsh-client-ui-conversation": "workspace:^", + "@deepseek-ai/dsh-client-ui-slots": "workspace:^", + "@deepseek-ai/dsh-client-web-react": "workspace:^", + "@deepseek-ai/dsh-invariants": "workspace:^", + "@deepseek-ai/dsh-plan-mode": "workspace:^", + "@types/react": "~18.3.1", + "cordis": "^4.0.0-rc.7", + "react": "^18.2.0" + }, + "files": [ + "lib/index.js", + "lib/invariant.js", + "lib/client.js", + "lib/types/**/*.d.ts", + "lib/types/**/*.d.ts.map", + "src" + ] +} diff --git a/packages/client/ui-plan/src/client/PlanModeControl.module.css b/packages/client/ui-plan/src/client/PlanModeControl.module.css new file mode 100644 index 0000000000..4bf42bd2a1 --- /dev/null +++ b/packages/client/ui-plan/src/client/PlanModeControl.module.css @@ -0,0 +1,56 @@ +/* Read-only plan status badge: quiet chip; the × affordance appears on + hover/focus and the whole chip is the /plan off button. */ + +.wrap { + display: inline-flex; + align-items: center; + gap: 6px; +} + +.chip { + display: inline-flex; + align-items: center; + gap: 4px; + padding: 6px 8px; + border: none; + border-radius: 8px; + background: transparent; + color: var(--dsw-alias-label-secondary); + font-size: 14px; + line-height: 20px; + cursor: pointer; +} + +.chip:hover:not(:disabled) { + background: var(--dsw-alias-interactive-bg-hover); +} + +.chip:focus-visible { + outline: 2px solid var(--dsw-alias-label-secondary); + outline-offset: 2px; +} + +.chip:disabled { + opacity: 0.6; + cursor: default; +} + +.close { + display: inline-flex; + align-items: center; + color: var(--dsw-alias-label-caption); + opacity: 0; + transition: opacity 0.12s ease; +} + +.chip:hover .close, +.chip:focus-visible .close { + opacity: 1; + color: var(--dsw-alias-label-secondary); +} + +.error { + color: var(--dsw-alias-state-error-primary); + font-size: 12px; + line-height: 18px; +} diff --git a/packages/client/ui-plan/src/client/PlanModeControl.tsx b/packages/client/ui-plan/src/client/PlanModeControl.tsx new file mode 100644 index 0000000000..b7945789a0 --- /dev/null +++ b/packages/client/ui-plan/src/client/PlanModeControl.tsx @@ -0,0 +1,74 @@ +import { useEffect, useRef, useState } from 'react' +import type { InjectFace, PropsRuntime } from '@deepseek-ai/dsh-client-ui-slots' +// Type-only: pulls the ui-conversation SlotMap merge (the input.plan seat and +// its {locked} owner share). +import type {} from '@deepseek-ai/dsh-client-ui-conversation/client' +import type { PlanChipInjected } from './index.ts' +import css from './PlanModeControl.module.css' + +/** Full plan-seat component props: runtime share (standard kit + locked owner prop) & injected share. */ +export type PlanChipProps = + PropsRuntime<'conversation.input.plan'> & InjectFace + +/** + * Read-only status badge over the host-computed `plan` projection. Plan mode + * is entered through the /plan command only; the chip appears while the + * effective target is plan mode and its hover × executes /plan off. The + * displayed state follows the target (`pending ? !active : active`) — a + * folded host value, not client optimism, so an arriving frame corrects it. + */ +export function PlanChip({ useProjection, locked, exitPlanMode }: PlanChipProps) { + const plan = useProjection('plan') + const [leaving, setLeaving] = useState(false) + const [error, setError] = useState(null) + const aliveRef = useRef(true) + + useEffect(() => { + aliveRef.current = true + return () => { + aliveRef.current = false + } + }, []) + + // Absent capability (no plan-mode host plugin / no session yet) or the + // default mode: no seat content. + if (plan === undefined) return null + const target = plan.pending ? !plan.active : plan.active + if (!target) return null + + const off = (): void => { + // No leaving/locked guard: both disable the button, so no click arrives. + setLeaving(true) + setError(null) + void exitPlanMode().then((failure) => { + if (!aliveRef.current) return + setLeaving(false) + setError(failure) + }, (reason: unknown) => { + if (!aliveRef.current) return + setLeaving(false) + setError(reason instanceof Error ? reason.message : String(reason)) + }) + } + + return ( + + + {error !== null && 退出 plan mode 失败} + + ) +} diff --git a/packages/client/ui-plan/src/client/index.ts b/packages/client/ui-plan/src/client/index.ts new file mode 100644 index 0000000000..2780d66a09 --- /dev/null +++ b/packages/client/ui-plan/src/client/index.ts @@ -0,0 +1,50 @@ +/** + * Plan control plugin, browser half: occupies the composer's named + * `conversation.input.plan` seat with a read-only status chip. Plan mode is + * entered through the /plan command only; while the projection's effective + * target is plan mode the chip renders (hover × executes /plan off through + * `command.execute`), otherwise the seat stays empty. Reads ride the generic + * projection pair through the standard-kit `useProjection` (an absent key is + * capability absence); zero client-side plan state. + */ +import type { ConnectionHandle } from '@deepseek-ai/dsh-client-connection/client' +import type { ClientContext, SessionId } from '@deepseek-ai/dsh-client-runtime/client' +// Type-only: pulls the ui-conversation SlotMap merge (the input.plan seat). +import type {} from '@deepseek-ai/dsh-client-ui-conversation/client' +// Type-only: pulls the `plan` SessionProjectionMap merge for useProjection. +import type {} from '@deepseek-ai/dsh-plan-mode/client' +import { PlanChip } from './PlanModeControl.tsx' + +/** Injected business face of the composer plan seat. */ +export interface PlanChipInjected { + /** + * Leave plan mode by executing /plan off. + * @returns null on admitted execution; a user-visible failure line otherwise. + */ + exitPlanMode: () => Promise +} + +/** + * Required services: the seat's slot registry, the transport, and the + * conversation service whose presence guarantees the seat is declared. + */ +export const inject = ['slots', 'connection', 'conversation'] + +/** + * Client plugin body: register the plan chip over the command channel. + * @param ctx - client root context. + */ +export function apply(ctx: ClientContext): void { + ctx.effect(() => ctx.slots.register({ + name: 'conversation.input.plan', + inject: (sessionId: SessionId): PlanChipInjected => ({ + exitPlanMode: async () => { + const connection = ctx.get('connection') as ConnectionHandle + const { result } = await connection.api.commands.execute({ sessionId, line: '/plan off' }) + if (!result.ok) return `${result.error.message}（${result.error.code}）` + if (!result.value.matched) return '未知命令：/plan off' + return null + }, + }), + }, PlanChip), 'ui-plan: composer plan chip registration') +} diff --git a/packages/client/ui-plan/src/css-modules.d.ts b/packages/client/ui-plan/src/css-modules.d.ts new file mode 100644 index 0000000000..8811db1264 --- /dev/null +++ b/packages/client/ui-plan/src/css-modules.d.ts @@ -0,0 +1,4 @@ +declare module '*.module.css' { + const classes: Record + export default classes +} diff --git a/packages/client/ui-plan/src/index.ts b/packages/client/ui-plan/src/index.ts new file mode 100644 index 0000000000..8d63cb2102 --- /dev/null +++ b/packages/client/ui-plan/src/index.ts @@ -0,0 +1,11 @@ +/** + * Plan control plugin, node half. Pure UI plugin: the empty apply exists so + * the plugin appears in the host cordis.yml / Loader; the browser half ships + * via exports["./client"], discovered through the package.json dshClient + * declaration. Plan behavior itself (the /plan command, the plan projection + * unit, the policy section) is owned by `@deepseek-ai/dsh-plan-mode`, + * composed independently on the host roster. + */ + +/** Host plugin body — no host-side behavior for this surface plugin. */ +export function apply(): void {} diff --git a/packages/client/ui-plan/src/invariant.ts b/packages/client/ui-plan/src/invariant.ts new file mode 100644 index 0000000000..82c9fc9376 --- /dev/null +++ b/packages/client/ui-plan/src/invariant.ts @@ -0,0 +1,31 @@ +/** + * Package-owned invariant companion for `@deepseek-ai/dsh-client-ui-plan`. + * @module @deepseek-ai/dsh-client-ui-plan/invariant + */ + +/* jscpd:ignore-start */ +import type { Context } from 'cordis' +import type { InvariantInstaller } from '@deepseek-ai/dsh-invariants' + +const PACKAGE_NAME = '@deepseek-ai/dsh-client-ui-plan' + +/** Cordis companion plugin name. */ +export const name = 'client-ui-plan-invariant' +/** Service required before the companion can reserve package ownership. */ +export const inject = ['invariants'] + +/** + * No runtime invariant: plan state and boundary ownership are + * audited by dsh-plan-mode, while the control is a slot effect whose + * declaration, registration, and teardown are exercised by this package. + */ +const install: InvariantInstaller = () => {} + +/** + * Register this package's invariant companion. + * @param ctx - Cordis context carrying the invariant service. + * @returns The installed registration's disposer after setup succeeds. + */ +export const apply = (ctx: Context): Promise<() => void> => + Promise.resolve(ctx.invariants.register(PACKAGE_NAME, install)) +/* jscpd:ignore-end */ diff --git a/packages/client/ui-plan/tests/browser-plugin.spec.ts b/packages/client/ui-plan/tests/browser-plugin.spec.ts new file mode 100644 index 0000000000..865ff67042 --- /dev/null +++ b/packages/client/ui-plan/tests/browser-plugin.spec.ts @@ -0,0 +1,78 @@ +/** + * ui-plan browser half on a real SlotsService: the plugin occupies the + * conversation-declared `conversation.input.plan` single seat with the plan + * status chip; the injected face executes /plan off and folds admission + * outcomes into null (admitted) or a user-visible failure line; teardown + * empties the seat (HMR safety). + */ +import { Context } from 'cordis' +import { describe, expect, it, vi } from 'vitest' +import { SlotsService } from '@deepseek-ai/dsh-client-runtime/client' +import type { SessionId } from '@deepseek-ai/dsh-client-runtime/client' +import { PlanChip } from '../src/client/PlanModeControl.tsx' +import type { PlanChipInjected } from '../src/client/index.ts' +import { apply, inject } from '../src/client/index.ts' +import { apply as nodeApply } from '../src/index.ts' + +const SID = 's-plan' as SessionId + +async function bench() { + const ctx = new Context() + await ctx.plugin(SlotsService).await() + const slots = ctx.get('slots') as SlotsService + slots.register({ + name: 'root', + children: { 'conversation.input.plan': { kind: 'single', scope: 'session' } }, + } as never, () => null) + const execute = vi.fn((_payload: { sessionId: SessionId; line: string }) => + Promise.resolve({ result: { ok: true as const, value: { matched: true as const, commandId: 'c1' } } })) + ctx.provide('connection', { api: { commands: { execute } } }) + ctx.provide('conversation', {}) + return { ctx, slots, execute } +} + +describe('ui-plan browser apply', () => { + it('declares every service it binds', () => { + expect(inject).toEqual(['slots', 'connection', 'conversation']) + }) + + it('node-half apply is an intentional no-op', () => { + expect(() => { nodeApply() }).not.toThrow() + }) + + it('fails loud when conversation did not declare the plan seat', async () => { + const ctx = new Context() + await ctx.plugin(SlotsService).await() + ctx.provide('connection', {}) + ctx.provide('conversation', {}) + await expect(ctx.plugin({ inject: [...inject], apply })) + .rejects.toThrow(/slot "conversation.input.plan" is not declared/) + }) + + it('registers the chip, executes /plan off, and unregisters on teardown', async () => { + const b = await bench() + const fiber = b.ctx.plugin({ inject: [...inject], apply }) + await fiber.await() + const entry = b.slots.entries('conversation.input.plan')[0]! + expect(entry.component).toBe(PlanChip) + const injected = (entry.inject as unknown as (id: SessionId) => PlanChipInjected)(SID) + + await expect(injected.exitPlanMode()).resolves.toBeNull() + expect(b.execute).toHaveBeenLastCalledWith({ sessionId: SID, line: '/plan off' }) + + // Business failure folds to the composer-visible line. + b.execute.mockResolvedValueOnce({ + result: { ok: false as const, error: { code: 'session-not-found', message: 'gone', details: {} } }, + } as never) + await expect(injected.exitPlanMode()).resolves.toBe('gone（session-not-found）') + + // Unmatched admission (plan-mode not composed host-side) is also a failure line. + b.execute.mockResolvedValueOnce({ + result: { ok: true as const, value: { matched: false as const } }, + } as never) + await expect(injected.exitPlanMode()).resolves.toBe('未知命令：/plan off') + + await fiber.dispose() + expect(b.slots.entries('conversation.input.plan')).toHaveLength(0) + }) +}) diff --git a/packages/client/ui-plan/tests/plan-mode-control.spec.tsx b/packages/client/ui-plan/tests/plan-mode-control.spec.tsx new file mode 100644 index 0000000000..44c17e5837 --- /dev/null +++ b/packages/client/ui-plan/tests/plan-mode-control.spec.tsx @@ -0,0 +1,111 @@ +// @vitest-environment jsdom +/** + * PlanChip over the `plan` projection: nothing renders while the capability + * is absent or the effective target is the default mode; the chip renders + * while the target is plan mode (pending follows the target — /plan shows it + * immediately, /plan off hides it immediately); the chip button executes + * /plan off and surfaces failures without hiding until the projection says so. + */ +import { afterEach, describe, expect, it, vi } from 'vitest' +import { cleanup, fireEvent, render, screen, waitFor } from '@testing-library/react' +import { createSnapshotStore } from '@deepseek-ai/dsh-client-runtime/client' +import { bindSnapshotSelector } from '@deepseek-ai/dsh-client-web-react' +import type { PlanProjection } from '@deepseek-ai/dsh-plan-mode/client' +import { PlanChip, type PlanChipProps } from '../src/client/PlanModeControl.tsx' + +afterEach(cleanup) + +function setup( + plan: PlanProjection | undefined, + exitPlanMode = vi.fn(() => Promise.resolve(null)), + locked = false, +) { + const store = createSnapshotStore<{ value: PlanProjection | undefined }>({ value: plan }) + const useProjection = (_key: string, selector?: (v: unknown) => unknown) => + bindSnapshotSelector(store)(s => (selector ?? (v => v))(s.value)) + const props = { useProjection, locked, exitPlanMode } as unknown as PlanChipProps + const view = render() + return { store, exitPlanMode, view } +} + +const chip = () => screen.getByRole('button', { name: 'Plan mode on, press to turn off' }) + +describe('PlanChip', () => { + it('renders nothing for absent capability or the default mode', () => { + const absent = setup(undefined) + expect(absent.view.container.innerHTML).toBe('') + cleanup() + const inactive = setup({ active: false, pending: false }) + expect(inactive.view.container.innerHTML).toBe('') + cleanup() + // Active with a pending exit: the target is default — chip already gone. + const leaving = setup({ active: true, pending: true }) + expect(leaving.view.container.innerHTML).toBe('') + }) + + it('renders while the effective target is plan mode, including the pending entry window', () => { + setup({ active: true, pending: false }) + expect(chip()).toBeTruthy() + cleanup() + // /plan just ran (command/run folded, plan/mode not yet): target is plan. + setup({ active: false, pending: true }) + expect(chip()).toBeTruthy() + }) + + it('the chip executes /plan off once and follows the projection down', async () => { + let resolve!: (value: string | null) => void + const exitPlanMode = vi.fn(() => new Promise((done) => { resolve = done })) + const { store } = setup({ active: true, pending: false }, exitPlanMode) + fireEvent.click(chip()) + expect(exitPlanMode).toHaveBeenCalledTimes(1) + // Busy while its own call is in flight. + fireEvent.click(chip()) + expect(exitPlanMode).toHaveBeenCalledTimes(1) + resolve(null) + // The off command's run record folds: target flips, the chip unmounts. + store.set({ value: { active: true, pending: true } }) + await waitFor(() => { + expect(screen.queryByRole('button', { name: 'Plan mode on, press to turn off' })).toBeNull() + }) + }) + + it('disables under the locked owner prop', () => { + setup({ active: true, pending: false }, vi.fn(), true) + expect((chip() as HTMLButtonElement).disabled).toBe(true) + }) + + it('surfaces admission and transport failures while staying visible', async () => { + const exitPlanMode = vi.fn() + .mockResolvedValueOnce('host said no') + .mockRejectedValueOnce(new Error('network down')) + .mockRejectedValueOnce('socket closed') + setup({ active: true, pending: false }, exitPlanMode) + fireEvent.click(chip()) + expect((await screen.findByText('退出 plan mode 失败')).getAttribute('title')).toBe('host said no') + expect(chip()).toBeTruthy() + + fireEvent.click(chip()) + expect(await screen.findByTitle('network down')).toBeTruthy() + + fireEvent.click(chip()) + expect(await screen.findByTitle('socket closed')).toBeTruthy() + }) + + it('ignores in-flight fulfillment and rejection after unmount', () => { + let resolve!: (value: string | null) => void + const successful = setup( + { active: true, pending: false }, + vi.fn(() => new Promise((done) => { resolve = done })), + ) + fireEvent.click(chip()) + successful.view.unmount() + expect(() => { resolve(null) }).not.toThrow() + + let reject!: (reason: unknown) => void + const exitPlanMode = vi.fn(() => new Promise((_done, fail) => { reject = fail })) + const { view } = setup({ active: true, pending: false }, exitPlanMode) + fireEvent.click(chip()) + view.unmount() + expect(() => { reject(new Error('late')) }).not.toThrow() + }) +}) diff --git a/packages/client/ui-plan/tsconfig.json b/packages/client/ui-plan/tsconfig.json new file mode 100644 index 0000000000..4ab13662f2 --- /dev/null +++ b/packages/client/ui-plan/tsconfig.json @@ -0,0 +1,36 @@ +{ + "extends": "../../../tsconfig.base.client.json", + "compilerOptions": { + "rootDir": "src", + "outDir": "lib/types" + }, + "include": [ + "src" + ], + "references": [ + { + "path": "../../../vendor/cordis" + }, + { + "path": "../runtime" + }, + { + "path": "../connection" + }, + { + "path": "../ui-conversation" + }, + { + "path": "../ui-slots" + }, + { + "path": "../web-react" + }, + { + "path": "../../plan/plan-mode" + }, + { + "path": "../../support/invariants" + } + ] +} diff --git a/packages/client/ui-plan/tsdown.config.ts b/packages/client/ui-plan/tsdown.config.ts new file mode 100644 index 0000000000..7f6a91d3f6 --- /dev/null +++ b/packages/client/ui-plan/tsdown.config.ts @@ -0,0 +1,3 @@ +import { clientBundle } from '../tsdown.client.ts' + +export default clientBundle('@deepseek-ai/dsh-client-ui-plan', ['lib/types/index.js', 'lib/types/invariant.js']) diff --git a/packages/client/ui-primitives/src/icons/index.tsx b/packages/client/ui-primitives/src/icons/index.tsx index 23b3bcb835..8f5c333778 100644 --- a/packages/client/ui-primitives/src/icons/index.tsx +++ b/packages/client/ui-primitives/src/icons/index.tsx @@ -680,3 +680,14 @@ export const IconListPenOutline16 = ({ size = 16, className }: IconProps) => ( /> ) + +/** sparkle_16 (Others tool-row / goal strip leading glyph; hand-authored three-star + * approximation — the figma 43:31850 glyph is an SF Symbols "sparkles" text glyph, + * not extractable as vector data) */ +export const IconSparkle16 = ({ size = 16, className }: IconProps) => ( + + + + + +) diff --git a/packages/client/ui-primitives/tests/icons.spec.tsx b/packages/client/ui-primitives/tests/icons.spec.tsx index c06a2e52fc..487fbb696e 100644 --- a/packages/client/ui-primitives/tests/icons.spec.tsx +++ b/packages/client/ui-primitives/tests/icons.spec.tsx @@ -14,8 +14,8 @@ const icons = Object.fromEntries( const iconNames = Object.keys(icons) describe('ic_ds_ icon set', () => { - it('exports the full P-I set (43 deepsuite + 13 figma extracts)', () => { - expect(iconNames.length).toBe(56) + it('exports the full P-I set (43 deepsuite + 13 figma extracts + the hand-authored sparkle)', () => { + expect(iconNames.length).toBe(57) }) it.each(iconNames)('%s renders an svg with currentColor fills and no hardcoded palette', (name) => { diff --git a/packages/client/ui-question/README.i18n.yaml b/packages/client/ui-question/README.i18n.yaml index 476bd3f0b6..a58cd055a7 100644 --- a/packages/client/ui-question/README.i18n.yaml +++ b/packages/client/ui-question/README.i18n.yaml @@ -1,6 +1,6 @@ # Bilingual-pair consistency record (docs/i18n/README.md): the git blob hash of each # side as of the last confirmed-consistent state. Both languages carry equal authority; # after editing either side, bring the other along and re-record with: -# pnpm run verify-translation-pairing --write -README.md: 28132d1d0643f5e8f658ab77de9017467da2d172 -README.zh.md: c70e77fc90eb5b226ceabd8a1e7cc7ca6c011c40 +# pnpm run verify-translation-pairing --write packages/client/ui-question/README.md +README.md: 3a3cd639fc2834685230aca7c8087583e0a48c71 +README.zh.md: 1330578577da7ed7d0890595f675fd272fd5ebc7 diff --git a/packages/client/ui-question/README.md b/packages/client/ui-question/README.md index 28132d1d06..3a3cd639fc 100644 --- a/packages/client/ui-question/README.md +++ b/packages/client/ui-question/README.md @@ -4,7 +4,7 @@ English | [中文](README.zh.md) Web `ask_user_question` feature plugin. Its host half mounts `dsh-tool-ask-user` only when the Web feature is selected; its browser half registers the `question` entry in the conversation-owned `conversation.composer` keyed slot. -The component renders one question at a time with progress navigation, single- and multi-select choices, recommendation badges derived from label suffixes, and custom answers. Single-select choices advance immediately, and Enter submits once every question is answered or skipped; Enter during IME composition confirms the input candidate without advancing. It submits one structured answer batch for the whole request: “Skip this question” retains other drafts and emits the existing blank `{ selected: [] }` shape for that item, while close rejects the whole wait as `ASK_CANCELLED`. +The component renders one question at a time with progress navigation, single- and multi-select choices, recommendation badges derived from label suffixes, and custom answers. Question detail reuses the assistant-output `MarkdownText` primitive, including its GFM rendering and untrusted-content policy. The capped card keeps its title, navigation, and submission actions fixed while long detail and choices share an internal scroll region. Single-select choices advance immediately, and Enter submits once every question is answered or skipped; Enter during IME composition confirms the input candidate without advancing. It submits one structured answer batch for the whole request: “Skip this question” retains other drafts and emits the existing blank `{ selected: [] }` shape for that item, while close rejects the whole wait as `ASK_CANCELLED`. Selection state is local to a component keyed by the request rpcId. A replay with the same id preserves a still-mounted draft, while `question/resolved` from the host removes the composer. The host remains authoritative: successful HTTP delivery does not remove pending state locally. diff --git a/packages/client/ui-question/README.zh.md b/packages/client/ui-question/README.zh.md index c70e77fc90..1330578577 100644 --- a/packages/client/ui-question/README.zh.md +++ b/packages/client/ui-question/README.zh.md @@ -4,7 +4,7 @@ Web `ask_user_question` 功能插件。只有选择 Web 功能时，其主机侧才会挂载 `dsh-tool-ask-user`；浏览器侧会把 `question` 配置项注册到会话拥有的 `conversation.composer` 键控 slot 中。 -组件每次渲染一个问题，提供进度导航、单选和多选选项、由标签后缀派生的推荐徽标，以及自定义答案。单选选项会立即前进；所有问题均已回答或跳过后，Enter 会提交；IME 输入法组合期间按 Enter 只会确认输入候选，不会前进。组件为整个请求提交一批结构化答案：「跳过此问题」会保留其他草稿，并为该项发出既有的空 `{ selected: [] }` 形状；关闭则以 `ASK_CANCELLED` 拒绝整个等待。 +组件每次渲染一个问题，提供进度导航、单选和多选选项、由标签后缀派生的推荐徽标，以及自定义答案。问题详情复用助手输出的 `MarkdownText` 原语，包括其 GFM 渲染与不受信内容策略。封顶卡片保持标题、导航与提交动作固定，超长的详情与选项共享内部滚动区。单选选项会立即前进；所有问题均已回答或跳过后，Enter 会提交；IME 输入法组合期间按 Enter 只会确认输入候选，不会前进。组件为整个请求提交一批结构化答案：「跳过此问题」会保留其他草稿，并为该项发出既有的空 `{ selected: [] }` 形状；关闭则以 `ASK_CANCELLED` 拒绝整个等待。 选择状态只存在于以请求 rpcId 为 key 的组件本地。使用相同 id 回放时，只要组件仍挂载，就会保留草稿；主机发出的 `question/resolved` 则会移除编辑器。主机仍具有最终决定权：HTTP 交付成功不会在本地移除待处理状态。 diff --git a/packages/client/ui-question/src/client/QuestionComposer.module.css b/packages/client/ui-question/src/client/QuestionComposer.module.css index be4fd5791b..16dc982e36 100644 --- a/packages/client/ui-question/src/client/QuestionComposer.module.css +++ b/packages/client/ui-question/src/client/QuestionComposer.module.css @@ -74,11 +74,7 @@ } .detail { - margin: 2px 0 0; - color: var(--dsw-alias-label-tertiary); - font-size: 13px; - line-height: 20px; - font-weight: 400; + margin: 0 2px 8px; } .headerActions, @@ -120,13 +116,19 @@ cursor: default; } +.body { + display: flex; + flex: 1 1 auto; + flex-direction: column; + min-height: 0; + overflow-y: auto; + overscroll-behavior: contain; +} + .options { display: flex; flex-direction: column; gap: 4px; - /* The scrollable region of the capped card (ChatView list pattern). */ - min-height: 0; - overflow-y: auto; } .option { diff --git a/packages/client/ui-question/src/client/QuestionComposer.tsx b/packages/client/ui-question/src/client/QuestionComposer.tsx index 4380ae8db0..542a24b935 100644 --- a/packages/client/ui-question/src/client/QuestionComposer.tsx +++ b/packages/client/ui-question/src/client/QuestionComposer.tsx @@ -2,7 +2,7 @@ import { useMemo, useState, type KeyboardEvent } from 'react' import clsx from 'clsx' import { Button, IconCheckOutline16, IconChevronLeftOutline14, IconChevronRightOutline14, - IconCloseOutline16, IconEditOutline16, + IconCloseOutline16, IconEditOutline16, MarkdownText, } from '@deepseek-ai/dsh-client-ui-primitives' import { PendingQuestion, type QuestionAnswer, type QuestionComposerProps } from './contract/slots.ts' import css from './QuestionComposer.module.css' @@ -176,7 +176,6 @@ function QuestionFlow({ pending }: { pending: PendingQuestion }) { : question.question} {question.multiSelect === true && 可多选} - {question.detail !== undefined &&

{question.detail}

}
{index + 1} / {questions.length} @@ -204,79 +203,84 @@ function QuestionFlow({ pending }: { pending: PendingQuestion }) {
-
- {(question.options ?? []).map((option, optionIndex) => { - const selected = draft.selected.includes(option.label) - const display = parseRecommendedLabel(option.label) - return ( - - ) - })} - -
- {hasOptions && ( - - )} - {draft.customOpen && ( -