docs: rescan rebased documentation hierarchy

This commit is contained in:
Turtle
2026-08-05 16:16:57 +08:00
parent 012bb0a549
commit 8af3babaea
304 changed files with 1091 additions and 1983 deletions
+2 -2
View File
@@ -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/user/develop/basic/config.md
config.md: 26d2d48ebede74194fbf306aa97d214bdb99b722
config.zh.md: 1ca727b56b661df315026902d48fbcaf507fe08b
config.md: 11a2311464789f74537cc7c4435f83ec07ca26fd
config.zh.md: 4e827ecafa6bfaf87c3e3f118425e656e1254787
+8 -21
View File
@@ -31,13 +31,15 @@ export function apply(ctx: Context, config: Config) {
}
```
Configure it in `cordis.yml`:
Add the configuration to the inserted local plugin row in `scratch-plugin/cordis.yml`:
```yaml
- name: './src/my-plugin.ts'
config:
greeting: 'Hi there'
maxRetries: 5
- insert:
- id: hello
name: './src/my-plugin.ts'
config:
greeting: 'Hi there'
maxRetries: 5
```
When loading the plugin, Cordis uses the exported schema to validate configuration and fill defaults. Do not export a plain object as `Config`; it does not implement the Standard Schema interface required by Cordis.
@@ -91,22 +93,7 @@ The test is whether `cordis.yml` can change the value without a code edit.
### Fail loudly on invalid configuration
If configuration refers to an unregistered LLM provider route or another nonexistent resource, fail early instead of silently skipping it:
```ts
import type { Context } from 'cordis'
import type {} from '@deepseek-ai/dsh-llm'
export interface ModelConfig {
provider: string
}
export function apply(ctx: Context, config: ModelConfig) {
if (!ctx.llm.listProviders().some(provider => provider.id === config.provider)) {
throw new Error(`LLM provider "${config.provider}" is not registered`)
}
}
```
Express self-contained constraints in the schema so invalid configuration fails while the plugin loads. References to services or registered resources require dependency injection; the [services tutorial](../framework/service.md) introduces that contract.
## Work with HMR
+8 -21
View File
@@ -31,13 +31,15 @@ export function apply(ctx: Context, config: Config) {
}
```
用户`cordis.yml` 中这样使用
`scratch-plugin/cordis.yml` 新插入的本地插件行中添加配置
```yaml
- name: './src/my-plugin.ts'
config:
greeting: 'Hi there'
maxRetries: 5
- insert:
- id: hello
name: './src/my-plugin.ts'
config:
greeting: 'Hi there'
maxRetries: 5
```
插件加载时,Cordis 会通过导出的 schema 校验配置,并填充未提供字段的默认值。不要导出普通对象作为 `Config`,因为它不满足 Cordis 要求的 Standard Schema 接口。
@@ -91,22 +93,7 @@ export interface Config {
### 配置错误要响亮
如果配置引用了未注册的 LLM 提供方路由或其他不存在的资源,应该尽早报错,而不是静默跳过:
```ts
import type { Context } from 'cordis'
import type {} from '@deepseek-ai/dsh-llm'
export interface ModelConfig {
provider: string
}
export function apply(ctx: Context, config: ModelConfig) {
if (!ctx.llm.listProviders().some(provider => provider.id === config.provider)) {
throw new Error(`LLM provider "${config.provider}" is not registered`)
}
}
```
在 schema 中表达自身完备的约束,使无效配置在插件加载时失败。对服务或已注册资源的引用需要依赖注入;[服务教程](../framework/service.md)会介绍这项契约。
## 配合 HMR
+2 -2
View File
@@ -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/user/develop/basic/index.md
index.md: 5a9f8dfb8f2d87dfbd2ba30b4d09d002ae9b635c
index.zh.md: 6f8926aeac5dcf667e4d311505ef7b2a321659c2
index.md: 45c8dfe495cd99da46a7b259b407af8deff570b3
index.zh.md: 9d8ee47e6f07fb4a897f49487cb7327904573c1b
+21 -6
View File
@@ -2,7 +2,15 @@
English | [中文](index.zh.md)
This guide creates a minimal Harness plugin and loads it into an agent.
This tutorial creates a minimal Harness plugin and loads it into the Web UI. Start from a repository checkout that has completed the [quick start](../../guide/quickstart.md).
## Create a local project
From the repository root, create a scratch project for the tutorial:
```sh
mkdir -p scratch-plugin/src
```
## What is a plugin?
@@ -22,7 +30,7 @@ That is the complete shape.
## Create the plugin file
Create `src/my-plugin.ts` in your project:
Create `scratch-plugin/src/my-plugin.ts`:
```ts
import type { Context } from 'cordis'
@@ -37,14 +45,21 @@ export function apply(ctx: Context) {
## Register it in cordis.yml
Add an entry to `cordis.yml`:
Create `scratch-plugin/cordis.yml` as a Web overlay that inserts the local plugin:
```yaml
- id: hello
name: './src/my-plugin.ts'
- insert:
- id: hello
name: './src/my-plugin.ts'
```
After startup, the console prints `[hello-plugin] plugin loaded!`.
Start the Web UI with that overlay:
```sh
pnpm run dsh web --config ./scratch-plugin/cordis.yml
```
Open `http://127.0.0.1:3080`. The terminal prints `[hello-plugin] plugin loaded!` during startup.
## Automatic cleanup
+21 -35
View File
@@ -2,7 +2,15 @@
[English](index.md) | 中文
文带你编写一个最小的 Harness 插件并加载到 agent(智能体)中
教程会创建一个最小的 Harness 插件,并将其加载到 Web UI 中。请从已完成[快速开始](../../guide/quickstart.md)的仓库检出开始
## 创建本地项目
在仓库根目录创建本教程使用的临时项目:
```sh
mkdir -p scratch-plugin/src
```
## 插件是什么
@@ -22,7 +30,7 @@ export function apply(ctx: Context) {
## 创建插件文件
在你的项目目录下创建 `src/my-plugin.ts`
创建 `scratch-plugin/src/my-plugin.ts`
```ts
import type { Context } from 'cordis'
@@ -37,14 +45,21 @@ export function apply(ctx: Context) {
## 注册到 cordis.yml
在你的 `cordis.yml` 中添加一条
创建 `scratch-plugin/cordis.yml`,作为插入本地插件的 Web 覆盖层
```yaml
- id: hello
name: './src/my-plugin.ts'
- insert:
- id: hello
name: './src/my-plugin.ts'
```
启动后你会在控制台看到 `[hello-plugin] plugin loaded!`
使用该覆盖层启动 Web UI
```sh
pnpm run dsh web --config ./scratch-plugin/cordis.yml
```
打开 `http://127.0.0.1:3080`。启动期间,终端会打印 `[hello-plugin] plugin loaded!`
## 自动清理
@@ -120,35 +135,6 @@ export default class MyService extends Service {
大多数情况下,函数形式足够了。当插件需要向其他插件提供服务时,可使用类形式(见 [服务与依赖](../framework/service.md))。
## 完整示例
最小的工具插件会在 `ctx.tools` 上注册其定义:
```ts
import type { Context } from 'cordis'
import { defineTool } from '@deepseek-ai/dsh-tools'
export const name = 'greet-tool'
export const inject = ['tools']
export function apply(ctx: Context) {
ctx.tools.register(defineTool({
name: 'greet',
description: 'Greet the named person.',
parameters: {
name: { type: 'string', required: true },
},
output: {
schema: { type: 'string' },
render: (_args, value) => [{ type: 'text', text: value }],
},
async execute(args) {
return `Hello, ${args.name}!`
},
}))
}
```
## 下一步
- [开发一个工具](./tool.md) — 详细了解工具定义 DSL
+2 -2
View File
@@ -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/user/develop/basic/tool.md
tool.md: 0d7cbc3f0b86f88fb67aeff6aa61181dff2912ee
tool.zh.md: 37362c510a07fcc32e9eee2b578d82907788e471
tool.md: 93a1a96feba814a564f8800c8e7b865fe9c0cb73
tool.zh.md: 18e6b9b5d9c17b00c26aa7b98e6ac4531315dc5e
+14 -207
View File
@@ -2,15 +2,17 @@
English | [中文](tool.zh.md)
A tool is a capability the model can call. This guide builds one with `defineTool`.
This tutorial adds a `greet` tool to the Web UI. Complete [Your first plugin](./) first and keep its `scratch-plugin` directory.
## Minimal example
## Create the tool plugin
Replace `scratch-plugin/src/my-plugin.ts` with:
```ts
import type { Context } from 'cordis'
import { defineTool } from '@deepseek-ai/dsh-tools'
export const name = 'my-tool'
export const name = 'greet-tool'
export const inject = ['tools']
export function apply(ctx: Context) {
@@ -25,221 +27,26 @@ export function apply(ctx: Context) {
render: (_args, value) => [{ type: 'text', text: value }],
},
async execute(args) {
// args is inferred as { name: string }.
return `Hello, ${args.name}!`
},
}))
}
```
## Parameter definitions
`inject` makes Cordis wait for the tool registry. `defineTool` infers and validates `args` from `parameters`; `execute` returns the canonical value declared by `output.schema`, and `output.render` converts that value to model-facing content.
`parameters` uses a compact format that the framework converts to the JSON Schema sent to the model.
## Run and call the tool
### Primitive types
Restart the development command if it is not running:
```ts
export const parameters = {
path: { type: 'string', required: true },
limit: { type: 'integer' },
recursive: { type: 'boolean' },
parent: { type: 'null' },
}
// Inferred type: { path: string; limit?: number; recursive?: boolean; parent?: null }
```sh
pnpm run dsh web --config ./scratch-plugin/cordis.yml
```
### Enums
```ts
export const parameters = {
mode: { type: 'string', required: true, enum: ['read', 'write', 'append'] },
}
// Inferred type: { mode: 'read' | 'write' | 'append' }
```
### Nested objects
```ts
export const parameters = {
options: {
type: 'object',
additionalProperties: true,
properties: {
timeout: { type: 'number' },
retries: { type: 'number' },
},
},
}
// The declared fields are inferred; additional JSON-valued keys are allowed.
```
### Arrays
```ts
export const parameters = {
tags: {
type: 'array',
items: { type: 'string' },
},
}
// Inferred type: { tags?: string[] }
```
### Property fields
| Field | Type | Meaning |
|------|------|------|
| `type` | `'string' \| 'number' \| 'integer' \| 'boolean' \| 'null' \| 'object' \| 'array' \| 'json'` | Value type; `json` accepts any lossless JSON value |
| `required` | `true` | Marks the property required and affects inference |
| `description` | `string` | Description sent to the model |
| `enum` / `const` | matching scalar values | Allowed literal values, checked at author and runtime boundaries |
| `properties` | `ParameterSchemaSpec` | Nested properties for an object |
| `additionalProperties` | `true \| false` | Required on every explicit object node |
| `items` | `ValueSchemaSpec` | Element schema for an array |
| `oneOf` | at least two `ValueSchemaSpec` branches | Requires exactly one matching branch; used instead of `type` |
The outer `parameters` map is an implicit open object. Explicit nested objects choose their openness; raw JSON Schema registered without `defineTool` keeps JSON Schema's open-by-default behavior.
## The execute function
`execute` receives validated, inferred `args` and an `exec` execution context:
```ts
import { defineTool } from '@deepseek-ai/dsh-tools'
export const tool = defineTool({
name: 'example',
description: 'Return an example result.',
parameters: {},
output: {
schema: { type: 'string' },
render: (_args, value) => [{ type: 'text', text: value }],
},
async execute(args, exec) {
// args: inferred from parameters
// exec: ToolExecution context
// Return the value declared by output.schema.
void args
void exec
return 'result here'
},
})
```
### Return value
`execute` returns the lossless JSON value declared by `output.schema`. `output.render(args, value)` separately turns that validated value into the Native/model-facing content:
```ts ignore-check
output: {
schema: {
type: 'object',
additionalProperties: false,
properties: {
path: { type: 'string', required: true },
content: { type: 'string', required: true },
},
},
render: (_args, value) => [{ type: 'text', text: value.content }],
},
async execute(args) {
return { path: args.path, content: await readFile(args.path, 'utf8') }
}
```
The canonical value is available to execution-time programmatic callers and is not persisted in `tool/result`; the rendered content and optional `presentationMeta` are the replayable projections. A body value that does not satisfy the schema, or is not lossless JSON, becomes an `INVALID_TOOL_OUTPUT` failure.
### Argument validation
Before calling `execute`, `defineTool` validates model-generated arguments. Invalid input raises `ToolArgsError`; the framework turns it into an `isError` result so the model can correct its call.
Do not repeat type validation inside `execute`.
## Presentation
A tool can define transport-neutral presentation methods for terminal and web clients:
```ts ignore-check
defineTool({
name: 'bash',
// ...
output: {
schema: { type: 'string' },
render: (_args, value) => [{ type: 'text', text: value }],
},
presentCall(args) {
return {
card: 'terminal',
title: args.command,
}
},
presentResult(args, result) {
return {
card: 'terminal',
output: result.content.map(b => b.type === 'text' ? b.text : '').join(''),
}
},
})
```
`presentCall` and `presentResult` are **pure functions**. Streaming UI and session replay may call them more than once.
## Registration and unloading
`ctx.tools.register()` returns a disposer, but a registration made through `ctx` is already tracked by the framework. Unloading the plugin removes the tool automatically, so the plugin does not call the disposer itself.
```ts ignore-check
// This is sufficient:
ctx.tools.register(defineTool({ /* ... */ }))
// No saved disposer or extra cleanup registration is needed.
```
## Complete example
This tool counts files in a directory:
```ts
import type { Context } from 'cordis'
import { defineTool } from '@deepseek-ai/dsh-tools'
import { readdir } from 'node:fs/promises'
export const name = 'file-counter'
export const inject = ['tools']
export function apply(ctx: Context) {
ctx.tools.register(defineTool({
name: 'count_files',
description: 'Count files in a directory.',
parameters: {
path: { type: 'string', required: true, description: 'Directory path' },
extension: { type: 'string', description: 'Filter by extension (e.g. ".ts")' },
},
output: {
schema: {
type: 'object',
additionalProperties: false,
properties: {
count: { type: 'integer', required: true },
files: { type: 'array', required: true, items: { type: 'string' } },
},
},
render: (_args, value) => [{ type: 'text', text: `Found ${value.count} files.` }],
},
async execute(args) {
const entries = await readdir(args.path, { withFileTypes: true })
let files = entries.filter(e => e.isFile())
if (args.extension) {
files = files.filter(f => f.name.endsWith(args.extension!))
}
return { count: files.length, files: files.map(file => file.name) }
},
}))
}
```
Open `http://127.0.0.1:3080` and ask: `Use the greet tool to greet Ada.` The model can call `greet` and receives `Hello, Ada!` as the tool result.
## Next steps
- [Plugin configuration](./config.md) — make the tool configurable
- [Capability layering](../practice/) — understand the interface/implementation/consumer pattern
- [Plugin configuration](./config.md) — make the greeting configurable.
- [Tool authoring reference](../../../cookbook/adding-a-tool.md) — look up nested schemas, canonical values, background work, policy hooks, Code Mode, and UI cards.
- [Capability layering](../practice/) — split a replaceable capability into interface, implementation, and consumer packages.
+15 -208
View File
@@ -1,16 +1,18 @@
# 开发一个工具
# 构建工具
[English](tool.md) | 中文
工具是模型可以调用的能力。本文介绍如何用 `defineTool` 编写一个工具
本教程会在 Web UI 中添加一个 `greet` 工具。请先完成[第一个插件](./),并保留其中的 `scratch-plugin` 目录
## 最小示例
## 创建工具插件
`scratch-plugin/src/my-plugin.ts` 替换为:
```ts
import type { Context } from 'cordis'
import { defineTool } from '@deepseek-ai/dsh-tools'
export const name = 'my-tool'
export const name = 'greet-tool'
export const inject = ['tools']
export function apply(ctx: Context) {
@@ -25,221 +27,26 @@ export function apply(ctx: Context) {
render: (_args, value) => [{ type: 'text', text: value }],
},
async execute(args) {
// args is inferred as { name: string }.
return `Hello, ${args.name}!`
},
}))
}
```
## 参数定义
`inject` 让 Cordis 等待工具注册表就绪。`defineTool` 根据 `parameters` 推导并校验 `args``execute` 返回 `output.schema` 声明的规范值,`output.render` 再将该值转换为面向模型的内容。
`parameters` 用一种简洁的格式描述参数,框架会自动转换为模型需要的 JSON Schema。
## 运行并调用工具
### 基本类型
如果开发命令未在运行,请重新启动:
```ts
export const parameters = {
path: { type: 'string', required: true },
limit: { type: 'integer' },
recursive: { type: 'boolean' },
parent: { type: 'null' },
}
// Inferred type: { path: string; limit?: number; recursive?: boolean; parent?: null }
```sh
pnpm run dsh web --config ./scratch-plugin/cordis.yml
```
### 枚举
```ts
export const parameters = {
mode: { type: 'string', required: true, enum: ['read', 'write', 'append'] },
}
// Inferred type: { mode: 'read' | 'write' | 'append' }
```
### 嵌套对象
```ts
export const parameters = {
options: {
type: 'object',
additionalProperties: true,
properties: {
timeout: { type: 'number' },
retries: { type: 'number' },
},
},
}
// The declared fields are inferred; additional JSON-valued keys are allowed.
```
### 数组
```ts
export const parameters = {
tags: {
type: 'array',
items: { type: 'string' },
},
}
// Inferred type: { tags?: string[] }
```
### 每个属性的字段
| 字段 | 类型 | 说明 |
|------|------|------|
| `type` | `'string' \| 'number' \| 'integer' \| 'boolean' \| 'null' \| 'object' \| 'array' \| 'json'` | 值类型;`json` 接受任意无损 JSON 值 |
| `required` | `true` | 标记为必填(影响类型推导) |
| `description` | `string` | 发送给模型的描述 |
| `enum` / `const` | 匹配类型的标量值 | 允许的字面量值,在编写和运行时边界校验 |
| `properties` | `ParameterSchemaSpec` | 对象的嵌套属性 |
| `additionalProperties` | `true \| false` | 每个显式对象节点都必须声明 |
| `items` | `ValueSchemaSpec` | 数组的元素 schema |
| `oneOf` | 至少两个 `ValueSchemaSpec` 分支 | 要求恰好匹配一个分支;代替 `type` 使用 |
外层 `parameters` 映射是一个隐式的开放对象。显式嵌套对象需自行选择是否开放;不通过 `defineTool` 注册的原始 JSON Schema 保持 JSON Schema 的默认开放语义。
## execute 函数
`execute` 接收经过校验的 `args`(类型自动推导)和一个 `exec` 上下文对象:
```ts
import { defineTool } from '@deepseek-ai/dsh-tools'
export const tool = defineTool({
name: 'example',
description: 'Return an example result.',
parameters: {},
output: {
schema: { type: 'string' },
render: (_args, value) => [{ type: 'text', text: value }],
},
async execute(args, exec) {
// args: inferred from parameters
// exec: ToolExecution context
// Return the value declared by output.schema.
void args
void exec
return 'result here'
},
})
```
### 返回值
`execute` 返回由 `output.schema` 声明的无损 JSON 值。`output.render(args, value)` 会将经过校验的值另外转换为 Native/模型可见的内容:
```ts ignore-check
output: {
schema: {
type: 'object',
additionalProperties: false,
properties: {
path: { type: 'string', required: true },
content: { type: 'string', required: true },
},
},
render: (_args, value) => [{ type: 'text', text: value.content }],
},
async execute(args) {
return { path: args.path, content: await readFile(args.path, 'utf8') }
}
```
执行期间的程序化调用方可以使用规范值,但 `tool/result` 不会持久化该值;渲染后的内容和可选的 `presentationMeta` 才是可回放的投影。工具主体返回的值若不满足 schema 或不是无损 JSON,就会变为 `INVALID_TOOL_OUTPUT` 失败。
### 参数校验
`defineTool` 在调用 `execute` 之前会自动校验模型生成的参数。如果参数不合法,会抛出 `ToolArgsError`,框架将其转换为 `isError` 结果返回给模型,让模型自行修正。
你不需要在 `execute` 里手动校验参数类型。
## 展示层(Presentation
工具可以定义与传输方式无关的展示方法,供终端和 Web 客户端使用:
```ts ignore-check
defineTool({
name: 'bash',
// ...
output: {
schema: { type: 'string' },
render: (_args, value) => [{ type: 'text', text: value }],
},
presentCall(args) {
return {
card: 'terminal',
title: args.command,
}
},
presentResult(args, result) {
return {
card: 'terminal',
output: result.content.map(b => b.type === 'text' ? b.text : '').join(''),
}
},
})
```
`presentCall` 和 `presentResult` 是**纯函数**,不能有副作用——UI 可能在流式传输中和会话回放中多次调用它们。
## 注册与卸载
`ctx.tools.register()` 返回值就是 disposer。但由于你在 `ctx` 上调用,框架已经自动追踪了这个注册——插件卸载时会自动移除工具。你不需要手动调用 disposer。
```ts ignore-check
// This is sufficient:
ctx.tools.register(defineTool({ /* ... */ }))
// No saved disposer or extra cleanup registration is needed.
```
## 完整示例
一个文件计数工具:
```ts
import type { Context } from 'cordis'
import { defineTool } from '@deepseek-ai/dsh-tools'
import { readdir } from 'node:fs/promises'
export const name = 'file-counter'
export const inject = ['tools']
export function apply(ctx: Context) {
ctx.tools.register(defineTool({
name: 'count_files',
description: 'Count files in a directory.',
parameters: {
path: { type: 'string', required: true, description: 'Directory path' },
extension: { type: 'string', description: 'Filter by extension (e.g. ".ts")' },
},
output: {
schema: {
type: 'object',
additionalProperties: false,
properties: {
count: { type: 'integer', required: true },
files: { type: 'array', required: true, items: { type: 'string' } },
},
},
render: (_args, value) => [{ type: 'text', text: `Found ${value.count} files.` }],
},
async execute(args) {
const entries = await readdir(args.path, { withFileTypes: true })
let files = entries.filter(e => e.isFile())
if (args.extension) {
files = files.filter(f => f.name.endsWith(args.extension!))
}
return { count: files.length, files: files.map(file => file.name) }
},
}))
}
```
打开 `http://127.0.0.1:3080`,然后输入:`Use the greet tool to greet Ada.` 模型可以调用 `greet`,并收到 `Hello, Ada!` 这一工具结果。
## 下一步
- [插件配置](./config.md) — 让你的工具可配置
- [能力分层](../practice/) — 了解接口/实现/消费方模式
- [插件配置](./config.md) — 让问候语可配置
- [工具编写参考](../../../cookbook/adding-a-tool.md) — 查阅嵌套 schema、规范值、后台工作、策略钩子、Code Mode 和 UI 卡片。
- [能力分层](../practice/) — 将可替换能力拆分为接口、实现和消费方包。