Two review findings (#220) on the sandbox context façade:
- Undeclared services were reachable: the façade resolved any live global via
ctx.get(name), so ctx.bash worked without inject: ['bash']. A cross-mount
consumer could then depend on a provider cordis never saw — unmounting the
provider would neither park the consumer nor unwind its registered tools,
leaving a model-visible tool that fails only at execution. The façade now
reads ctx.fiber.inject and refuses any service the mount did not declare
(with a teaching error naming the inject fix), so the dependency is always
visible to cordis and its activation/unload semantics bind.
- ctx.tools.get returned the live ToolDefinition, including execute — mount
code could call another tool directly and bypass ToolRegistry.execute and
its pre/post-execute hooks and accounting. get now returns the same
read-only name/description/parameters view as schemas(), never an invocable.
Adds inject-gate and schema-view regression cases to sandbox-context.spec.ts
(undeclared property/get denied, declared allowed, the cross-mount zombie-tool
scenario refused at call time, get exposes no execute). Package stays at
per-file 100% coverage. RFC, mount description, and tool-catalog updated.
Review finding (#220): the guarded proxy only special-cased ctx.tools, so
mount code could reach an UNGUARDED context through ctx.root, ctx.extend(), or
a service instance's .ctx, then ctx.root.tools.register({…}) to bypass the
marker check and host-realm normalization — a raw vm-realm result would later
error a real agent turn at the session-log plainness check.
The sandbox ctx is now a whitelist façade, not a pass-through proxy: it exposes
only what a mount needs — tools.register (marker-guarded), on/once, provide, the
timer helpers, and injected services resolved through a guarded get — and denies
every framework-plumbing member (root, parent, fiber, reflect, registry, extend,
isolate, intercept, plugin, set, mixin, …) with a teaching error. Injected
services are wrapped so a method returning a Context is rejected on the way back
(the .ctx escape), closing the one indirect leak. There is no context-valued
member left to reach; cross-mount provide/inject is untouched (the plugin's own
inject and the fiber's pending/active gating are unchanged). ctx.plugin (child
plugins) and ctx.set are denied by design; ctx.effect is deferred (FIXME).
Adds tests/sandbox-context.spec.ts covering the escape class (root/extend/fiber/
plugin/set/… denied, the classic root.tools.register bypass, the .ctx escape,
read-only writes) plus the async-service and symbol/in-operator paths for 100%
coverage. RFC/README/tool-catalog/config-catalog updated; api-catalog.ts
regenerated (also picks up the codeRuntime service that entered on the master
merge and was left stale).
The tree SHAPE was the wrong surface for the model: what it needs from
cordis_inspect is what services, plugins, and capabilities are loaded, not the
fiber hierarchy. The plugins section is now a flat name + lifecycle-state list
from ctx.registry (deterministically sorted, one line per instance); the ASCII
tree renderer, the parent→child rebuild, and the dyn-id tree annotation are
deleted — dynamic mounts keep their own richer dynamic section (id, state,
provides, waits). Net -49 lines; RFC and READMEs state the flat-list contract.
Field sessions showed models writing tool schemas in the JSON-Schema dialect
by strong prior — type: 'integer', required: false, then the full
{ type:'object', properties, required: [...] } wrapper — and the rejection
text itself pushed a nearly-correct DSL attempt BACK to raw JSON Schema: one
stats tool cost three consecutive schema errors before mounting. The boundary
now normalizes wherever the input has exactly one meaning (wrapper unwrapped
with the required array becoming per-property flags at any nesting level,
integer → number, required: false → optional, all rebuilt as fresh host-realm
objects) and rejects only genuinely meaningless input, enumerating the valid
vocabulary in the error. Re-running the failing session mounts first-try.
The mount description documents both accepted forms.
The design record for tool-cordis: the three-tool contract, the vm sandbox
trust stance and boundary mechanisms, the dynamic-group lifecycle, cross-mount
provide/inject composition, the generated runtime API catalog, and the
alternatives weighed (per-capability registration tools, hand-maintained API
tables, a mount provenance event, a hardened sandbox).
The RFC's deployment-policy decision is unchanged; state the current
mechanism in place — the per-tool budget is declared on ToolDefinition
(timeoutMs, set by the owning tool plugin from its config) and the
enforcer is zero-config, so a mistyped tool name is impossible.
Regenerate config-catalog (timeout-policy -> no-config; tool-web gains
fetch/searchTimeoutMs), the event graph (tools/change loses its
timeout-policy consumer), the ToolDefinition type-equiv block, and a
source-line drift in the cordis services catalog.
The seam's structured-clone boundary admits values JSON does not (BigInt,
Map, circulars), while tool/code-dispatch events must be JSON-appendable —
left unhandled, a sub-call could execute and then fail at logging time.
The bridge now JSON-normalizes binding arguments BEFORE dispatch (a value
that does not survive rejects that one call), so the dispatched form and
the logged form are the same JSON value by construction.
Model-facing tool-call budgets were tangled into each capability's schema
(bash timeoutMs, web_fetch timeout_ms) with no shared home. Add a
tools/execute around-dispatch waterfall to dsh-tools whose base next() is
the dispatch-with-normalization thunk, and a new @deepseek-ai/dsh-timeout-policy
plugin (packages/timeout/) that arms a per-tool deadline on exec.signal and
returns a structured TOOL_TIMEOUT when it wins. Migrate web_fetch (drop the
model-facing timeout_ms) and web_search onto it; the fetch provider keeps its
timeout only as a resource backstop for direct callers. bash and hook command
execution keep BASH_TIMEOUT unchanged.
Named the plugin timeout-policy (not the RFC's tool-timeout) so it does not
trip the gen-tool-catalog packages/*/tool-* completeness guard, and replace
exec.signal by in-place mutation before next() since cordis waterfall next()
ignores passed arguments. RFC moved to implemented/ recording both deviations.
A client-callback throw only becomes a JSON-RPC error RESPONSE to the
agent's session/request_permission — runScenario itself kept going, so a
tolerant agent could treat the error as a denial and the scenario would
pass, or worse, record: the impossible click baked into fixture and
golden, green on every replay. The mismatch is now captured as a harness
error while the agent is answered plain cancelled (a well-defined path
it cannot reinterpret), and the step loop rejects the run on it as soon
as the in-flight step settles. The spec asserts the rejection instead of
the agent-side error echo.
The package, coverage, and permission scripting all shipped on this
branch, so the RFC moves to implemented/ with the lifecycle rewrite:
Proposal becomes a present-tense Decision, Acceptance criteria and Risks
fold into Testing/Consequences with what actually pinned each one (the
zero-byte extraction parity, the 100% per-file coverage via the fake
bin, the vitest-in-src caveat, the per-suite pin cost).
The snapshot tier's machinery leaves examples/acp-agent/tests for
packages/support/acp-snapshot (@deepseek-ai/dsh-acp-snapshot), where the
coverage gate measures it and a second example can consume it instead of
forking it: harness.ts (runScenario, parameterized by an AgentUnderTest
{binScript, configPath, tsconfigPath} instead of module constants),
normalize.ts (moved verbatim), and suite.ts (defineAcpSnapshotSuite — the
per-scenario golden/log compares, record write-back, per-suite header pin
with its uniformity guard, and the fixture guard block, lifted from
acp.snapshot.ts). The example file collapses to its scenario table plus
one factory call; env reading (DSH_SNAPSHOT) stays at that edge.
The exactly-one-pin meta-test generalizes from the hardcoded text-turn
name to "exactly one per suite" — which scenario pins is the scenario
table's reviewable choice (per-suite pinning per the proposal RFC).
Extraction parity: pnpm run test:snapshot is 36 passed + fs-policy-reject
failing BEFORE AND AFTER (BSD-sed environment failure, reproduced at the
base commit in a clean worktree — the recorded golden's sed -i syntax is
GNU-only), with zero byte changes under examples/acp-agent/tests/snapshots/.
Coverage for the new src files lands in the next commit.
The harness, normalizers, and suite/guard logic live inside
examples/acp-agent/tests, outside the coverage gate and copyable-only
for a second suite. Propose @deepseek-ai/dsh-acp-snapshot under
packages/support: parameterized runScenario, verbatim normalizers, a
defineAcpSnapshotSuite factory with per-suite header pinning, and
scripted permissionAnswers so an approval round-trip is expressible at
the snapshot tier — the sandbox composition is the immediate consumer.
New group packages/code-runtime/ with the interface package
@deepseek-ai/dsh-code-runtime, per the Code Mode RFC: abstract CodeRuntime
service (run() resolves program failures as an error field, rejects only
for seam misuse), the CodeRunRequest/CodeBindingNamespace/CodeRunResult/
CodeLogEntry/CodeRunFailure vocabulary, and readonly language/isolation
backend descriptors. Registered in the tsconfig maps, packages/README,
architecture service map, and the doc-graph service-role classification;
catalogs regenerated.
The RFC's one forward path token to the worker package becomes an npm-name
mention until PR3 creates that directory (verify-package-paths is
drift-scoped: the now-existing group made the token checkable).
docs/architecture.md ceiling 1630 -> 1640: the doc gained a genuinely new
capability-service row; the row itself is already minimal.
Budget expiry terminated the worker but nothing cancelled an in-flight
host-side sub-dispatch, and a late dispatch could append events after
run_code returned. The bridge now owns a run-scoped AbortController
(follows exec.signal; fired on any run settlement), sub-dispatches get
the run signal, and run_code returns only after the dispatch queue
drains — no post-settlement appends, per dispose-to-quiescence.
(A1) Scope the wire-collapse guarantee honestly: systemPrompt.tools() is
a public multi-provider API, so the mode governs the registry's
contribution (the only shipped source); deliberate extra providers own
what they add, and the shipped-configuration invariant is test-pinned.
(A2) Replace pause-on-pending-RPC timeout with two independent budgets:
computeMs metered by worker.performance.eventLoopUtilization() busy time
(unfoolable by an un-awaited decoy dispatch; probe-verified) plus a
never-pausing maxWallMs ceiling.
(A3) Specify sub-call additionalContext as deliberately suppressed in
the MVP (immediate inject would break call/result adjacency; the plural
channel is named follow-up work).
(B) Orphan-process caveat vs bash-local's group kill; null-prototype
binding namespaces (__proto__/constructor names); per-PR doc artifacts
(packages/README row, architecture service map in PR2, config/tool/
persistence catalogs per owning PR); engines range corrected to
^22.19.0 || >=24.0.0.
Research finding: a SessionEventMap member is a log event — JSDoc prose
required, @mode is a hard error there, and docs/persistence-catalog.md
must be regenerated (todo/write is the log-only precedent). PR4's plan
now names both.
Own adversarial pass finding: a static inject on the registry would gate
ctx.tools (and every tool plugin) on a code runtime existing even under
mode 'native'. The RFC now names the sanctioned pattern: soft
ctx.get('codeRuntime') at use time (the agent-loop sessionPersistence
precedent) with absence failing loud in the provider thunk.
Supersedes the add-on-plugin + node:vm-stub draft in place (still proposed/):
code mode becomes a first-class ToolRegistry presentation mode
('native' | 'code' | 'both'), execution goes behind a new ctx.codeRuntime
capability seam whose shipped backend is one fresh Node worker thread per
run (type-strip, empty env, resource limits, hard terminate), at
bash-equivalent trust with no unsafe-flag ceremony. Renames the file to
2026-06-15-code-mode.md and regenerates the RFC index.
Review follow-up (#196): a listed name with no registered tool was silently
ignored; misconfiguration must block work instead. The check lives in the
assembly — the earliest moment the registered tool set exists (tool plugins
register after the service constructs) and the only universal one (cordis has
no "all plugins loaded" event; registrations change at any time). assemble()
is now async so the throw surfaces as a rejection rather than a synchronous
escape from a Promise-returning method.
Blast radius, pinned by a loop-level test: the rejection reaches the turn's
outer catch — the turn closes balanced with an `error` reason, agent/error
mirrors it, no step opens, no request/header is logged, no request reaches
the adapter, and the agent returns to idle; every turn fails identically
until the config is fixed. A boot-time validation pass was considered and
rejected (recorded in the RFC). The general principle — misconfiguration
fails loud, never a silent skip — is added to AGENTS.md.
A three-dot rest entry reads as elision in a cordis.yml; the spelled-out
sentinel says what lands there. The literal now appears once in code (the
constant) and once in the value-pinning test; every other reference — the
forwarding tests included — imports TOOL_ORDER_REST, which adds the
dsh-system-prompt devDependency to the two app packages.
Review follow-up on #196.
- run-gates.ts docSyncLeafGates() gains verify-export-jsdoc — CI lanes
and the pre-push hook execute this leaf list, not the doc-sync npm
script, so the gate was previously unenforced there (proven by
SessionForkErrorCode landing undocumented via a master merge while
checks stayed green; now documented). Same wiring gap fixed for
master's verify-config-catalog, which was also missing from the list.
- The heritage exemption now recovers the base's return surface: a void
base return carried no @returns duty, so an override returning a
concrete result documents it itself (annotated overrides run the
standard check; unannotated ones are classified by the checker so
faithful void overrides need no boilerplate annotation). Three new
negative-path tests pin it; RFC and module doc updated.
- AGENTS.md states each principle inline instead of citing RFCs (eight
citations removed; high-level doc links kept) and the editing section
now carries the self-containedness rule.
- Generated catalogs/graphs regenerated for the shifted line pointers.
Out-of-process delegation to external coding agents as two new subagent
seam backends, exposed as subagent_claude_code / subagent_codex tools.
Verified against @anthropic-ai/claude-agent-sdk 0.3.202 and codex CLI
0.142.5 via keyless spikes; includes the dsh-subagent-process extraction
plan, isolation/permission stances, and tiered test coverage.
imccyu
Dudu-0223
kingwl
Tianyi Cui
pku-xht
Hypatia May