dsh-plugin-dev-kb 1.4.0 → 1.5.2

package/kb/extra/cookbook/adding-a-settings-card.md

@@ -0,0 +1,72 @@
+# Cookbook: live configuration forms
+
+English | [中文](adding-a-settings-card.zh.md)
+
+Declare live fields in the plugin Config schema and expose them through a product-owned settings card. The exported `Config` interface describes the values received by the plugin, including each `Volatile<T>` reference.
+
+## 1. Declare live fields
+
+```ts
+import type { Context, Volatile } from '@deepseek-ai/cordis'
+import z from '@deepseek-ai/schemastery'
+
+export interface Config {
+  endpoint: Volatile<string | undefined>
+  retries: Volatile<number>
+}
+
+export const Config = z.object({
+  endpoint: z.string().volatile(),
+  retries: z.number().step(1).min(0).default(3).volatile(),
+})
+
+export function apply(ctx: Context, config: Config): void {
+  ctx.on('loader/volatile-update', () => {
+    ctx.logger.info('Retry limit: %d', config.retries.get())
+  })
+}
+```
+
+Read `.get()` when starting an operation. A request that needs a consistent snapshot captures its values once. Use `.check()` for cross-field Config validation; these checks run on the Host before persistence and are omitted from serialized form schemas.
+
+## 2. Compose the plugin
+
+Give each instance a unique profile entry id. The base bundle mounts settings and config-editor. For custom profiles, see their package READMEs before mounting the same services.
+
+`role('secret')` keeps a value out of form responses. Use credential references for values managed by the credentials domain. Ordinary fields stay out of the settings schema.
+
+## 3. Verify editing
+
+Change a field in Plugins and save. Verify the profile patch, the next consumer operation, unchanged plugin instance identity, and restoration after restart. Reject an invalid value and verify that neither the file nor the live value changed.
+
+Custom plugin pages receive `form.state` and `form.mutate(operations, expectedRevision)` from the Plugins page owner. The Host validates full Config and applies edits through ConfigEditor and volatile HMR. The [plugin settings package](../../packages/client/ui-settings-plugins/README.md) owns the existing card examples.
+
+## 4. Contribute to another plugin's page
+
+A plugin with something to say about a bundle, a row, or an official plugin it does not own registers into `plugins.detail.actions` (a control at the head of the page), `plugins.detail.badge` (a tag beside the title), or `plugins.detail.section` (a section under the page's own content). Every entry is rendered with the page's `subject` — `{ kind: 'bundle', pkg }`, `{ kind: 'row', pkg, row }`, or `{ kind: 'item', id }` — and returns null for a subject it has nothing for:
+
+```tsx ignore-check
+ctx.slots.inject('plugins.detail.badge', () => ctx.slots.register({
+  name: 'plugins.detail.badge',
+  id: 'acme-update',
+  locale: 'acmeUpdate',
+}, ({ t, subject }) => subject.kind === 'bundle' && hasUpdate(subject.pkg) ? <Tag tone="info">{t('update')}</Tag> : null))
+```
+
+## 5. Where the browser half rides
+
+The browser half is served to the page by the [client module system](../../packages/client/modules), which scans the enabled Loader entries for packages declaring `dsh.client` and serves each one's built `./client` export — but it attaches a package's half to the Loader row whose specifier is the bare package name. A row mounted from a subpath export never carries a half, so a bundle that splits one package into several rows keeps its half on the root row, and every page it registers goes away when that row is switched off. A sub-plugin whose page must outlive the other rows ships as its own package.
+
+The built `./client` file must be in the client module system's lazy-CJS factory format: one script that registers the package name and a `factory(require)` with the page's module loader, described in the [client module system's README](../../packages/client/modules/README.md). The `clientBundle` tsdown preset that emits it lives in `packages/client/tsdown.client.ts` rather than in a published package, so a package outside this repository reproduces that build itself.
+
+```jsonc
+{
+  "exports": {
+    ".": { "types": "./lib/types/index.d.ts", "default": "./lib/index.js" },
+    "./client": { "types": "./lib/types/client/index.d.ts", "default": "./lib/client.js" }
+  },
+  "dsh": { "client": { "platform": "web", "inject": ["@deepseek-ai/dsh-client-ui-settings"] } }
+}
+```
+
+A companion package for a built-in namespace is the same browser half in a client-only package with an empty Host `apply`, listed in the web composition's plugin roster ([`packages/bundle/web-app/cordis.patch.yml`](../../packages/bundle/web-app/cordis.patch.yml)) and registering into `plugins.item` through `ctx.configForms.whileServed`, so the page exists exactly while the Host serves the namespace.

package/kb/extra/cookbook/adding-a-settings-card.zh.md

@@ -0,0 +1,72 @@
+# 实践指南：即时配置表单
+
+[English](adding-a-settings-card.md) | 中文
+
+在插件 Config schema 中声明即时字段，并通过产品所属的设置卡片提供编辑入口。导出的 `Config` 接口描述插件收到的值，包括每个 `Volatile<T>` 引用。
+
+## 1. 声明即时字段
+
+```ts
+import type { Context, Volatile } from '@deepseek-ai/cordis'
+import z from '@deepseek-ai/schemastery'
+
+export interface Config {
+  endpoint: Volatile<string | undefined>
+  retries: Volatile<number>
+}
+
+export const Config = z.object({
+  endpoint: z.string().volatile(),
+  retries: z.number().step(1).min(0).default(3).volatile(),
+})
+
+export function apply(ctx: Context, config: Config): void {
+  ctx.on('loader/volatile-update', () => {
+    ctx.logger.info('Retry limit: %d', config.retries.get())
+  })
+}
+```
+
+在操作开始时读取 `.get()`。需要一致快照的请求一次性捕获所需值。使用 `.check()` 进行跨字段 Config 验证；这些检查在持久化前由 Host 执行，不进入序列化的表单 schema。
+
+## 2. 组合插件
+
+为每个实例分配唯一的 profile 条目 id。基础组合包挂载 settings 和 config-editor。自定义 profile 挂载这些服务前，请阅读相应包的 README。
+
+`role('secret')` 阻止值进入表单响应。对于凭据域管理的值，使用凭据引用。普通字段不进入设置 schema。
+
+## 3. 验证编辑
+
+在插件页面修改字段并保存。验证 profile patch、消费者下一次操作、插件实例标识不变，以及重启后的恢复。提交无效值，并确认文件和即时值均未改变。
+
+自定义插件页面从 Plugins 页面所有者接收 `form.state` 和 `form.mutate(operations, expectedRevision)`。Host 校验完整 Config，并通过 ConfigEditor 和 volatile HMR 应用修改。[插件设置包](../../packages/client/ui-settings-plugins/README.zh.md) 提供现有卡片示例。
+
+## 4. 向其他插件的页面贡献内容
+
+对某个不属于自己的组合包、行或官方插件有话要说的插件，注册进 `plugins.detail.actions`（页头的控件）、`plugins.detail.badge`（标题旁的标签）或 `plugins.detail.section`（页面自身内容之下的区块）。每个条目都以页面的 `subject` 渲染——`{ kind: 'bundle', pkg }`、`{ kind: 'row', pkg, row }` 或 `{ kind: 'item', id }`——对无话可说的 subject 返回 null：
+
+```tsx ignore-check
+ctx.slots.inject('plugins.detail.badge', () => ctx.slots.register({
+  name: 'plugins.detail.badge',
+  id: 'acme-update',
+  locale: 'acmeUpdate',
+}, ({ t, subject }) => subject.kind === 'bundle' && hasUpdate(subject.pkg) ? <Tag tone="info">{t('update')}</Tag> : null))
+```
+
+## 5. 浏览器半侧挂在哪里
+
+浏览器半侧由[客户端模块系统](../../packages/client/modules)送到页面：它扫描已启用的 Loader 条目，找出声明了 `dsh.client` 的包，送出每个包构建好的 `./client` 导出——但它只把一个包的半侧挂在说明符恰为裸包名的那一行上。从子路径导出挂载的行永远不带半侧，因此把一个包拆成多行的组合包，其半侧留在根行上，它注册的每个页面都随根行关闭而消失。需要在其他行关闭时仍保留页面的子插件，应作为独立的包发布。
+
+构建出的 `./client` 文件必须是客户端模块系统的 lazy-CJS factory 格式：一段脚本，向页面的模块加载器登记包名和一个 `factory(require)`，见[客户端模块系统的 README](../../packages/client/modules/README.zh.md)。生成它的 `clientBundle` tsdown 预设位于 `packages/client/tsdown.client.ts`，而不在任何已发布的包里，因此仓库之外的包要自己复刻这一步构建。
+
+```jsonc
+{
+  "exports": {
+    ".": { "types": "./lib/types/index.d.ts", "default": "./lib/index.js" },
+    "./client": { "types": "./lib/types/client/index.d.ts", "default": "./lib/client.js" }
+  },
+  "dsh": { "client": { "platform": "web", "inject": ["@deepseek-ai/dsh-client-ui-settings"] } }
+}
+```
+
+内置命名空间的伴生包就是同样的浏览器半侧放在一个纯客户端包里，宿主 `apply` 为空，列入 Web 组合的插件花名册（[`packages/bundle/web-app/cordis.patch.yml`](../../packages/bundle/web-app/cordis.patch.yml)），并通过 `ctx.configForms.whileServed` 注册进 `plugins.item`，页面因此恰好在 Host 服务该命名空间期间存在。

package/kb/extra/cookbook/adding-a-tool.md

@@ -0,0 +1,101 @@
+# Tool authoring reference
+
+English | [中文](adding-a-tool.zh.md)
+
+Reference for the contracts a model-facing tool must satisfy. For an ordered first tool, follow [Build a tool](../user/develop/basic/tool.md). `packages/shell/tool-bash` is the production-grade three-package example.
+
+## The minimal shape
+
+```ts
+import { readFile } from 'node:fs/promises'
+import type { Context } from '@deepseek-ai/cordis'
+import { defineTool } from '@deepseek-ai/dsh-tools'
+
+export const name = 'my-tool'
+export const inject = ['tools']
+
+export function apply(ctx: Context) {
+  ctx.tools.register(defineTool({
+    name: 'read_file',
+    description: 'Read a file from disk.',          // what the model sees
+    parameters: {
+      path: { type: 'string', required: true, description: 'Absolute path' },
+      limit: { type: 'number' },                     // optional by default
+    },
+    output: {
+      schema: { type: 'string' },
+      render: (_args, value) => [{ type: 'text', text: value }],
+    },
+    async execute(args, exec) {
+      // args is TYPED from the schema: { path: string; limit?: number }
+      // exec carries immutable identity + token; signal is the operational field
+      return readFile(args.path, { encoding: 'utf8', signal: exec.signal })
+    },
+  }))
+}
+```
+
+Registration is effect-based: disposing the plugin fiber unregisters the tool. Schemas flow into the system-prompt assembly automatically.
+
+## Rules of the execute() contract
+
+- **Args are validated for you.** `defineTool` validates model-generated `arguments` against the unified `ParameterSchemaSpec` before `execute` runs (types, required keys, literal constraints, exact-one unions, and nested values), so inside `execute` the args match `InferArgs`. Explicit object nodes declare `additionalProperties: true | false`; the implicit parameter root stays open. You still hand-check constraints the DSL does not express, such as non-empty strings, positive numbers, or cross-field rules. Raw JSON-Schema tools registered directly own their input validation.
+- **Registration borrows your readonly definition.** A typed same-process contribution is not a serialization boundary; do not mutate its schema or replace callbacks after registration. `schemas()` materializes only the explicit model-facing projection. To hot-swap a tool, dispose its owning effect and register the replacement; mutable state inside the callback's closure remains ordinary plugin state.
+- **Execution identity is protected.** The registry materializes `arguments` as detached lossless JSON in one recursive pass, freezes that value before policy starts, and assigns an opaque `exec.token`; `callId`, `name`, `arguments`, `agent`, `token`, the required caller-owned `signal`, and an optional enclosing-transport `parent` token stay immutable through dispatch. `parent` is identity-only and exposes no live outer execution. Treat `args` as readonly input. Only an around-dispatch wrapper receives a mutable view, and it may replace and restore the required `exec.signal` to impose a deadline but cannot remove it.
+- **Declare and return one canonical JSON value.** `output.schema` uses `ValueSchemaSpec` and may have an object, array, scalar, or null root. `execute` returns only the inferred value; the registry snapshots it as lossless JSON, validates it, freezes it, and passes it to `output.render(args, value)`. Do not return content blocks from the body or make callers parse prose for ids and fields.
+- **Throwing or returning an invalid value means `isError`.** The registry catches throws and contains schema, renderer, metadata-projector, and lossless-JSON failures before observers run. Throw for infrastructure failures. Represent a successful domain outcome in the canonical value even when its Native renderer explains a non-ideal state, such as a non-zero process exit.
+- **Honor `exec.signal`.** Cancel in-flight work when it fires.
+- **Project durable card data with `presentationMeta` (optional).** `output.presentationMeta(args, value)` derives replayable JSON from the same canonical value. The core persists it on `tool/result` and hands it to `presentResult`, so a card that needs result-time facts—such as `write`/`edit` applied hunks—survives replay without persisting the canonical value. The projector is skipped for nested Code dispatches because they have no cards.
+- **Use `exec.agent` for async notifications.** `agent.inject({ content, source: { kind: 'my-plugin' } })` appends durable context the NEXT model request sees — it is not a wake-up (an idle agent stays idle). Declare that `kind` through [`MessageSourceMap`](../subsystems/llm-streaming.md#content-blocks-and-messages) in your plugin: session format V4 rejects the retired `{ kind: 'plugin', plugin: '<name>' }` wrapper at message admission. Guard against disposed agents (try/catch).
+
+## Long-running work
+
+Gate `run_in_background` with producer config, then register through `ctx.jobs.start({ kind, label, owner: exec.agent, run })`. The registry rejects a pre-aborted invocation before the producer body; the runtime validates ownership and task-controller availability before `run()` starts work, then supplies the id, session fence, generic control tools, notices, and owner cleanup. A successful background branch returns a typed canonical handle such as `{ kind: 'background', jobId }`; its Native renderer may keep human prose such as `started background job bash-1`, but PTC mode must never parse that prose to recover the id.
+
+The spec supplies synchronous `cancel`, a non-rejecting `done` that settles after resource cleanup, and either pull `output` sources the registry pumps into the job's output ring or pushes through the `JobHandle` its starter receives; `dsh-tool-jobs` renders the model's consuming read from that ring. A pre-aborted call is a failure because no task exists whose id could satisfy the successful output schema. Once `ctx.jobs.start()` publishes the id, use a task-owned cancellation signal rather than `exec.signal`: later outer-call cancellation stops waiting for the call but does not kill published work; `job_kill`, owner disposal, and service teardown own that lifetime. Foreground work remains coupled to `exec.signal`. See the [background job runtime reference](../../packages/jobs/jobs/README.md) and `dsh-tool-bash` for a stream producer.
+
+## Execution policy and observation
+
+Prefer not to build deployment policy into the tool. Use `tools/pre-execute` for extensible allow/deny/ask policy (the [permission-gate example](extension-cookbook.md#a-hook-plugin-permission-gate-example)), `ctx.tools.guard()` for a final monotonic deny that later listeners cannot undo, `tools/execute` to wrap dispatch with a deadline, retry, or metrics collection, `tools/post-execute` to replace presentation content or the returned value, block the result, or attach model-facing context, and `tools/result` to observe the immutable normalized outcome. A content replacement leaves programmatic access to `value` intact; confidentiality policy blocks or replaces the value. A sandboxing implementation can also run inside the tool's executor implementation; the [`dsh-tools` README](../../packages/core/tools/README.md#extension-points) defines each extension point's inputs, order, return values, and failure behavior.
+
+## PTC mode reaches your tool for free
+
+In [PTC mode](../../packages/core/tools/README.md), every visible registered tool is available as `await tools.<name>(args)` without extra integration. The generated `ToolArgsMap` and `ToolOutputMap` derive exact argument and canonical-return types from the same schemas, and calls re-enter the normal execution pipeline. A successful call resolves to the final canonical JSON value after policy, not to rendered Native content. A failed call rejects with the real `ToolCallError`; programs can inspect only its `name`, `toolName`, and human-readable `message`, not internal error codes or a failure union.
+
+Design `output.schema` as a useful programmatic API: return handles and fields directly, allow scalar/array/null roots when they are the honest value, and keep human explanation in `output.render`. Intermediate values are execution-local, are not persisted or prompt-truncated, and have no byte cap, so the producer's truthful acquisition bounds and process memory still matter. Only the outer `run_code` logs/result cross the configurable output cap and model-facing spill pipeline.
+
+## How your tool renders in a UI
+
+Your tool's `output.render` returns model-facing content; its **UI card** is a separate concern declared through pure presentation projections and optional `presentCall` / `presentResult` methods. Design these alongside the canonical value. A tool with no UI presentation falls back to a generic card (title = tool name, raw args as input).
+
+Both methods return a **`card`-tagged render intent** — pick the card kind that matches what your tool does:
+
+- `presentCall(args)` → a `ToolCallView` (the PENDING card):
+  - `{ card: 'generic', title, kind?, rawInput?, content?, locations? }` — the default. Set `kind` for an icon (`read`/`search`/…); set `locations: [{ path, line? }]` for any file your tool touches so a capable editor follows along / jumps to it.
+  - `{ card: 'terminal', title, description?, cwd? }` — your call IS a shell command. `title` is the command, `description` renders above the terminal card. (tool-bash.)
+  - `{ card: 'diff', title, diffs, locations? }` — your call creates or modifies a file. `diffs: [{ path, oldText, newText }]` (`oldText: null` for a new file) renders as an inline diff card. (tool-fs `write`/`edit`.)
+- `presentResult(args, { content, isError, meta? })` returns the completed card:
+  - `generic` supplies an optional title and content.
+  - `terminal` supplies raw output and optional exit metadata; each UI renders its capable or fallback view.
+  - `diff` supplies applied hunks, often derived by `output.presentationMeta` and carried in persisted `result.meta` so replay reproduces them. Mutation tools keep a diff result because the completed view replaces the pending card.
+  - `read` supplies a completed file window reconstructed from persisted `result.meta`: the file `path`, a 1-based `offset`, the returned `lines` (each keeping its file line number), `totalLines`, and an optional `lang` highlight hint; a UI without the `read` capability falls back to the raw result content. There is no `read` call view — a read call's pending state stays a generic card, since content exists only after `execute`. (tool-fs `read`.)
+  - `search` supplies a discovery result reconstructed from persisted `result.meta`: grouped-by-file matches (`shape: 'matches'`, grep) or a flat path list (`shape: 'paths'`, glob), plus `truncated`/`total` so a UI never presents a capped result as complete. The view carries no result text (a UI without a search card falls back to the raw result content), and there is no `search` call view — a discovery call's pending state stays a generic card, since matches exist only after `execute`. (tool-fs-search `grep`/`glob`.)
+  - `web` supplies a completed web retrieval, discriminated by `kind: 'search' | 'fetch'` (the structured search sources or the fetch summary), derived from `result.meta`; it carries no body copy, so a UI without the `web` capability falls back to the raw result content. (tool-web `web_search`/`web_fetch`.)
+
+Hard rules (they bite if broken):
+
+- **Purity.** These run on live streaming AND on session-log REPLAY, so they must be pure functions of `args` (+ the result) — NO I/O, NO reading session state, NO clock/random. A diff is derived from the args (`write` uses `oldText: null` because a call-time presenter has no prior file content); the UI adapter, not the tool, supplies session context. If you find yourself wanting the file's old content or the working directory inside `presentCall`, stop — that belongs in durable result metadata or the adapter, not the presenter.
+- **UI-only formatting stays out of the model result.** A fenced ` ```console ` block, a diff, a relativized path—none of these belongs in the canonical value or Native content merely to serve a UI. `output.render` owns model-facing prose; `presentationMeta` plus the card presenters own replayable UI state. A `terminal` result view carries raw output and the adapter adds any fallback framing.
+- **`defineTool` soft-validates the display path.** Malformed or older logged arguments make the wrapper return `undefined` (a generic fallback) rather than throw — display must never crash a replay.
+
+The neutral vocabulary lives in `dsh-tools`; tools never import a UI or transport type. Consumers of this API map each `card` into their own view. The design and the why are in [the render-intent-union Agent Note](../../.agents/notes/implemented/architecture/2026-07-02-tool-render-intent-union.md); `dsh-tool-fs` (generic/diff) and `dsh-tool-bash` (terminal) are the reference implementations.
+
+## Web Client presentation
+
+The built-in Web Client does not consume `presentCall` or `presentResult`. Session `page` and `follow` transport raw `tool/call` and `tool/result` events, including persisted `result.meta`. A Client plugin registers its wire tool name in the `tool.call.toolview` keyed slot and derives component props from the `ToolCallBlock` arguments, content, error, metadata, existing PTC dispatch `parentCallId`, and Session path facts. It validates these wire values locally and returns the generic row for malformed or unsupported input.
+
+Use `output.presentationMeta(args, value)` when an existing Web card needs bounded structured result facts that model-facing content cannot preserve losslessly. Do not store React props or a selected card in metadata, import a Host tool implementation into a browser bundle, or create another Client presenter registry. Defining Host presentation methods alone does not add a specialized Web card. The [Client-derived presentation Agent Note](../../.agents/notes/implemented/architecture/2026-08-23-client-derived-tool-presentation.md) defines ownership, fallback, and equivalence requirements.
+
+## Verification
+
+Follow the [repository testing policy](../testing.md) and the owning package's test documentation. A shipped model- or UI-visible change requires the assembled coverage specified there.

package/kb/extra/cookbook/adding-a-tool.zh.md

@@ -0,0 +1,103 @@
+# 工具编写参考
+
+[English](adding-a-tool.md) | 中文
+
+面向模型的工具必须满足哪些约定，均以本文为准。如需按步骤构建第一个工具，请阅读[构建工具](../user/develop/basic/tool.zh.md)。`packages/shell/tool-bash` 是生产级的三包示例。
+
+## 最小形态
+
+```ts
+import { readFile } from 'node:fs/promises'
+import type { Context } from '@deepseek-ai/cordis'
+import { defineTool } from '@deepseek-ai/dsh-tools'
+
+export const name = 'my-tool'
+export const inject = ['tools']
+
+export function apply(ctx: Context) {
+  ctx.tools.register(defineTool({
+    name: 'read_file',
+    description: 'Read a file from disk.',          // what the model sees
+    parameters: {
+      path: { type: 'string', required: true, description: 'Absolute path' },
+      limit: { type: 'number' },                     // optional by default
+    },
+    output: {
+      schema: { type: 'string' },
+      render: (_args, value) => [{ type: 'text', text: value }],
+    },
+    async execute(args, exec) {
+      // args is TYPED from the schema: { path: string; limit?: number }
+      // exec carries immutable identity + token; signal is the operational field
+      return readFile(args.path, { encoding: 'utf8', signal: exec.signal })
+    },
+  }))
+}
+```
+
+注册基于副作用：dispose（资源释放）插件 fiber 即注销该工具。schema 会自动流入系统提示词的组装过程。
+
+## execute() 约定的规则
+
+- **参数已为你校验。** `defineTool` 在 `execute` 运行前，会根据统一的 `ParameterSchemaSpec` 校验模型生成的 `arguments`（类型、必填键、字面量约束、恰好匹配一个分支的联合以及嵌套值），因此 `execute` 内的 args 会匹配 `InferArgs`。显式对象节点必须声明 `additionalProperties: true | false`；隐式参数根对象保持开放。你仍需手动检查 schema DSL 无法表达的约束，例如非空字符串、正数或跨字段规则。直接注册的原始 JSON Schema 工具自行负责输入校验。
+- **注册借用你的只读定义。** 类型化的同进程贡献不是序列化边界；注册后不要修改其 schema 或替换回调。`schemas()` 只物化显式的模型可见投影。如需热替换工具，请 dispose 其所属副作用并注册替代品；回调闭包内的可变状态仍是普通的插件状态。
+- **执行身份受保护。** 注册表在一次递归遍历中将 `arguments` 物化为分离的无损 JSON，在策略开始前冻结该值，并分配一个不透明的 `exec.token`；`callId`、`name`、`arguments`、`agent`、`token`、必填且由调用方持有的 `signal`，以及可选的外层传输 `parent` token 在整个分发过程中保持不可变。`parent` 仅用于身份标识，不暴露活跃的外层执行。请将 `args` 视为只读输入。只有 around-dispatch 包装器会收到可变视图；它可以替换并恢复必填的 `exec.signal` 以施加截止时间，但不能移除该信号。
+- **声明并返回一个规范 JSON 值。** `output.schema` 使用 `ValueSchemaSpec`，根可以是对象、数组、标量或 null。`execute` 只返回推导出的值；注册表将其快照为无损 JSON，完成校验和冻结后，再传给 `output.render(args, value)`。工具主体不要返回内容块，也不要迫使调用方从自然语言中解析 id 和字段。
+- **抛出异常或返回无效值意味着 `isError`。** 注册表会捕获异常，并在观察者运行前收敛 schema、渲染器、元数据投影器和无损 JSON 失败。基础设施故障请抛异常。成功的领域结果即使表示不理想的状态，也应写入规范值；其 Native 渲染器可以解释该状态，例如进程以非零状态退出。
+- **遵守 `exec.signal`。** 信号触发时取消进行中的工作。
+- **使用 `presentationMeta` 投影持久化的卡片数据（可选）。** `output.presentationMeta(args, value)` 从同一个规范值派生可回放的 JSON。核心将其持久化在 `tool/result` 上并传给 `presentResult`，因此需要结果期事实的卡片——例如 `write`／`edit` 的已应用 hunk——无需持久化规范值也能在回放中重现。嵌套 Code 分发没有卡片，因此会跳过该投影器。
+- **使用 `exec.agent` 发送异步通知。** `agent.inject({ content, source: { kind: 'my-plugin' } })` 追加持久化上下文，下一次模型请求会看到它——这不是唤醒（空闲的 agent（智能体）保持空闲）。在插件里通过 [`MessageSourceMap`](../subsystems/llm-streaming.zh.md#content-blocks-and-messages) 声明该 `kind`：session format V4 在消息准入处拒绝退役的 `{ kind: 'plugin', plugin: '<name>' }` 包装。请防范已 dispose 的 agent（try/catch）。
+
+## 长时间运行的工作
+
+通过 producer 配置控制 `run_in_background`，然后使用 `ctx.jobs.start({ kind, label, owner: exec.agent, run })` 注册任务。注册表会在进入 producer 主体前将已预先中止的调用判为失败；运行时会在 `run()` 启动工作前校验 owner 和任务控制器是否可用，随后提供 id、会话围栏、通用控制工具、通知和 owner cleanup。成功的后台分支会返回类型化的规范句柄，如 `{ kind: 'background', jobId }`；其 Native 渲染器可以保留 `started background job bash-1` 这类供人阅读的自然语言，但 PTC mode 绝不能通过解析该文本取得 id。
+
+spec 提供同步的 `cancel`、在资源清理后 settle 且不 reject 的 `done`，以及由注册表泵入 job 输出环的拉取式 `output` 源或经 starter 收到的 `JobHandle` 推送的追加；模型的消费式读取由 `dsh-tool-jobs` 从该环渲染。预先中止的调用属于失败，因为此时没有任务，其 id 无法满足成功输出 schema。`ctx.jobs.start()` 发布 id 后，应使用任务自有的取消信号，而不是 `exec.signal`：之后取消外层调用只会停止等待本次调用，不会终止已经发布的工作；该生命周期归 `job_kill`、owner dispose 和服务 teardown 所有。前台工作仍与 `exec.signal` 耦合。流式 producer 的示例和完整约定见[后台任务运行时 reference](../../packages/jobs/jobs/README.zh.md)与 `dsh-tool-bash`。
+
+<a id="execution-policy-and-observation"></a>
+
+## 执行策略与观测
+
+尽量不要把部署策略内建到工具中。使用 `tools/pre-execute` 实现可扩展的允许／拒绝／询问策略（见[权限门禁示例](extension-cookbook.zh.md#a-hook-plugin-permission-gate-example)）；使用 `ctx.tools.guard()` 设置最终的单调拒绝，后续监听器无法撤销；使用 `tools/execute` 为分发添加截止时间、重试或指标收集；使用 `tools/post-execute` 替换展示内容或返回值、阻止结果，或附加模型可见上下文；使用 `tools/result` 观测不可变的归一化结果而不改变它。替换内容不会阻止程序化访问 `value`；保密策略会屏蔽或替换该值。沙箱实现也可以在工具的执行器实现中运行；[`dsh-tools` README](../../packages/core/tools/README.zh.md#extension-points) 定义每个扩展点的输入、顺序、返回值和失败行为。
+
+## PTC mode 自动触达你的工具
+
+在 [PTC mode](../../packages/core/tools/README.zh.md) 中，每个可见的已注册工具都可通过 `await tools.<name>(args)` 调用，无需额外集成。生成的 `ToolArgsMap` 和 `ToolOutputMap` 会根据同一组 schema 分别派生精确的参数类型与规范返回类型，调用则重新进入正常的执行流水线。成功调用会解析为策略处理后的最终规范 JSON 值，而不是渲染后的 Native 内容。失败调用会以真正的 `ToolCallError` reject；程序只能检查其 `name`、`toolName` 和可供人阅读的 `message`，无法取得内部错误代码或失败联合。
+
+请把 `output.schema` 设计为实用的程序化 API：直接返回句柄与字段；当标量、数组或 null 确实就是结果时，允许采用相应的根类型；将面向人类的解释放入 `output.render`。中间值只存在于执行期间，不会被持久化或按提示词上限截断，也不设字节上限，因此生产方如实声明的采集边界和进程内存仍然重要。只有外层 `run_code` 日志／结果会受到可配置输出上限和面向模型的 spill 流水线约束。
+
+## 工具在 UI 中的渲染方式
+
+工具的 `output.render` 返回模型可见的内容；其 **UI 卡片** 是另一项独立关注点，通过纯展示投影以及可选的 `presentCall`／`presentResult` 方法声明。请将这些内容与规范值一并设计。没有 UI 展示方法的工具会回退到通用卡片（标题 = 工具名，原始 args 作为输入）。
+
+两个方法都返回一个 **`card` 标签的渲染意图**——选择与你的工具行为匹配的卡片类型：
+
+- `presentCall(args)` → 一个 `ToolCallView`（PENDING 卡片）：
+  - `{ card: 'generic', title, kind?, rawInput?, content?, locations? }`——默认。设置 `kind` 获取图标（`read`／`search`／…）；设置 `locations: [{ path, line? }]` 标注工具涉及的文件，使有能力的编辑器跟随／跳转。
+  - `{ card: 'terminal', title, description?, cwd? }`——你的调用本身就是 shell 命令。`title` 是命令，`description` 渲染在终端卡片上方。（tool-bash。）
+  - `{ card: 'diff', title, diffs, locations? }`——你的调用创建或修改文件。`diffs: [{ path, oldText, newText }]`（新文件时 `oldText: null`）渲染为内联 diff 卡片。（tool-fs `write`／`edit`。）
+- `presentResult(args, { content, isError, meta? })` 返回完成后的卡片：
+  - `generic` 提供可选的标题和内容。
+  - `terminal` 提供原始输出和可选的退出元数据；各 UI 根据自身能力渲染对应视图或回退视图。
+  - `diff` 提供已应用的 hunk，通常由 `output.presentationMeta` 派生并通过持久化的 `result.meta` 携带，使回放能重现它们。变更类工具保留 diff 结果，因为完成后的视图会替换 pending 卡片。
+  - `read` 提供从持久化 `result.meta` 重建的已完成文件窗口：文件 `path`、从 1 开始的 `offset`、返回的 `lines`（每行保留其文件行号）、`totalLines`，以及可选的 `lang` 高亮提示；不具备 `read` 能力的 UI 回退到原始结果内容。没有 `read` 调用视图——读取调用的 pending 状态保持为 generic 卡片，因为内容只在 `execute` 之后才存在。（tool-fs `read`。）
+  - `search` 提供从持久化 `result.meta` 重建的发现型结果：按文件分组的匹配（`shape: 'matches'`，grep）或扁平路径列表（`shape: 'paths'`，glob），外加 `truncated`／`total` 使 UI 永不把被截断的结果当作完整结果呈现。该视图不携带结果文本（无 search 卡片的 UI 回退到原始结果内容），也没有 `search` 调用视图——发现型调用的 pending 状态保持为 generic 卡片，因为匹配只在 `execute` 之后才存在。（tool-fs-search 的 `grep`／`glob`。）
+  - `web` 提供已完成的 web 检索，以 `kind: 'search' | 'fetch'` 区分（结构化的搜索来源或抓取摘要），由 `result.meta` 派生；它不携带正文副本，因此不具备 `web` 能力的 UI 回退到原始结果内容。（tool-web `web_search`／`web_fetch`。）
+
+硬性规则（违反会出问题）：
+
+- **纯函数。** 这些方法在实时流式输出和会话日志回放时都会运行，因此必须是 `args`（加 result）的纯函数——不做 I/O、不读会话状态、不用时钟／随机数。diff 从 args 派生（`write` 使用 `oldText: null`，因为调用时的展示器没有文件先前内容）；会话上下文由 UI 适配器而非工具提供。如果你发现自己想在 `presentCall` 内获取文件旧内容或工作目录，请停下：那属于持久结果元数据或适配器，不属于展示器。
+- **UI 格式不进入模型结果。** 围栏 ` ```console ` 块、diff、相对化路径均不应仅为服务 UI 而进入规范值或 Native 内容。`output.render` 负责模型可见的自然语言；`presentationMeta` 和卡片展示器负责可回放的 UI 状态。`terminal` 结果视图携带原始输出，由适配器按需添加回退格式。
+- **`defineTool` 对展示路径做软校验。** 格式错误或旧版日志中的参数会使包装器返回 `undefined`（通用回退）而非抛异常——展示绝不能导致回放崩溃。
+
+中性词汇定义在 `dsh-tools` 中；工具绝不导入 UI 或传输类型。使用该 API 的消费方把每个 `card` 映射到自己的视图。设计与原因见[渲染意图联合体 Agent Note](../../.agents/notes/implemented/architecture/2026-07-02-tool-render-intent-union.zh.md)；`dsh-tool-fs`（generic/diff）和 `dsh-tool-bash`（terminal）是参考实现。
+
+## Web Client 展示
+
+内置 Web Client 不消费 `presentCall` 或 `presentResult`。Session `page` 与 `follow` 运输原始 `tool/call` 和 `tool/result` 事件，包括持久化的 `result.meta`。Client 插件在 keyed slot `tool.call.toolview` 中注册自己的 wire 工具名称，并从 `ToolCallBlock` 的参数、内容、错误、metadata、现有 PTC dispatch `parentCallId` 与 Session 路径事实派生组件 props。插件在本地校验这些 wire 值，并让格式错误或不受支持的输入回退到 generic 行。
+
+现有 Web 卡片需要模型可见内容无法无损保存的有界结构化结果事实时，使用 `output.presentationMeta(args, value)`。不要在 metadata 中保存 React props 或预选卡片，不要把 Host 工具实现导入浏览器 bundle，也不要建立另一套 Client presenter registry。只定义 Host 展示方法不会增加专用 Web 卡片。[Client 派生展示 Agent Note](../../.agents/notes/implemented/architecture/2026-08-23-client-derived-tool-presentation.zh.md)规定 owner、fallback 与对等要求。
+
+## 验证
+
+遵循[仓库测试策略](../testing.zh.md)和所属包的测试文档。已交付且面向模型或 UI 的变更必须提供其中规定的组装覆盖。

package/kb/extra/cookbook/adding-an-llm-adapter.md

@@ -0,0 +1,43 @@
+# Cookbook: adding an LLM adapter
+
+English | [中文](adding-an-llm-adapter.zh.md)
+
+How to connect a new model provider. Reference implementations: `packages/llm/llm-deepseek` (direct HTTP, SSE framed by `eventsource-parser`) and `packages/llm/llm-pi-ai` (wrapping an LLM library). Read the `StreamChunk` doc in `packages/llm/llm/src/types.ts` first — it records the protocol conventions both adapters were verified against.
+
+## The shape
+
+```ts ignore-check
+class MyAdapter extends LlmAdapter {
+  async * stream(options: GenerateOptions): AsyncIterable<StreamChunk> { … }
+}
+
+export const name = 'llm-myprovider'
+export const inject = ['llm']
+export const Config: z<Config> = z.object({ apiKey: z.string(), … })
+
+export function apply(ctx: Context, config: Config) {
+  ctx.llm.registerAdapter(['my-provider'], new MyAdapter(…))
+}
+```
+
+Registration is effect-based (HMR-safe); one adapter per provider route — duplicates throw, and multi-route registration is all-or-nothing. `options.provider` selects the adapter and `options.model` is the provider model id, so a dynamic catalog adapter can serve new models without lifecycle reconfiguration. Secrets are cordis-native: schemastery Config with env fallbacks, fed from cordis.yml via `!!js process.env.MY_KEY`. Never read ad-hoc key files in code.
+
+## Protocol obligations (the contract two implementations verified)
+
+- Emit `usage` BEFORE `finish`; emit NOTHING after `finish`. The robust way: buffer finish/usage until the provider's end-of-stream marker, then flush (handles providers that send trailing usage-only chunks).
+- Tool-call `arguments` are RAW JSON strings end-to-end; stream fragments as `argumentsDelta`. If your provider hands back parsed objects, re-stringify at `block-end`.
+- Allocate block `index`es in first-seen stream order; reuse the index for every delta of the same block.
+- Errors have exactly two sanctioned paths: THROW from `stream()` (transport and protocol failures — use `LlmError` with a stable code), or end the stream with `finish {kind: 'error' | 'aborted'}` (provider in-band failures). Consumers handle both; pick per failure class and document it.
+- Honor `options.signal` (pass it to fetch / your SDK).
+- A `GenerateOptions` field your provider cannot honor (e.g. a `stop` list on a provider without stop sequences): throw `LlmError(..., 'UNSUPPORTED_OPTION')` rather than silently dropping it.
+- If the provider requires response ids, signatures, or other native metadata on follow-up calls, emit the minimal lossless-JSON projection as `finish.replayState`. Validate it when rebuilding history. `LlmRuntime` passes it only when the historical provider route and target provider route are currently owned by the exact same adapter instance; your adapter decides whether same-model, cross-model, or cross-provider restoration is legal. Never infer native replay from provider/model names alone when state is absent.
+
+Provider-specific thinking-mode toggles remain in the adapter's Config. Exact model metadata uses one provider-neutral capability seam: implement `resolveModel()` with provider/model identity and optional `context` and `reasoning` fields, declare a configured `defaultEffort` only when one exists, and honor the resolver's optional `AbortSignal`. Reasoning efforts are ordered opaque ids mapped to provider requests by the adapter. Preserve the adapter's authoritative selectable list, including an adapter-defined `off` when supported, without exposing final wire spellings or clamping unsupported values; an id need not equal its wire representation.
+
+## Implementation structure
+
+Keep wire types, request serialization, transport parsing, chunk translation, and the adapter class as separate responsibilities; [`llm-deepseek`](../../packages/llm/llm-deepseek/README.md) is the reference layout.
+
+## Verification
+
+Follow the [repository testing policy](../testing.md), which owns adapter coverage, real-provider checks, and published-entry requirements.

package/kb/extra/cookbook/adding-an-llm-adapter.zh.md

@@ -0,0 +1,43 @@
+# 实操手册：添加 LLM（大语言模型）适配器
+
+[English](adding-an-llm-adapter.md) | 中文
+
+如何接入一个新的模型提供方。参考实现：`packages/llm/llm-deepseek`（直接 HTTP，SSE（Server-Sent Events）由 `eventsource-parser` 分帧）与 `packages/llm/llm-pi-ai`（封装 LLM 库）。请先阅读 `packages/llm/llm/src/types.ts` 中的 `StreamChunk` 文档——它记录了两个适配器都经过验证的协议约定。
+
+## 基本形态
+
+```ts ignore-check
+class MyAdapter extends LlmAdapter {
+  async * stream(options: GenerateOptions): AsyncIterable<StreamChunk> { … }
+}
+
+export const name = 'llm-myprovider'
+export const inject = ['llm']
+export const Config: z<Config> = z.object({ apiKey: z.string(), … })
+
+export function apply(ctx: Context, config: Config) {
+  ctx.llm.registerAdapter(['my-provider'], new MyAdapter(…))
+}
+```
+
+注册基于副作用，可安全支持 HMR（热模块替换）；每个提供方路由仅对应一个适配器，重复注册会抛出异常，多路由注册要么全部成功，要么全部失败。`options.provider` 用于选择适配器，`options.model` 是提供方模型 ID，因此动态模型目录适配器无需重新配置生命周期即可提供新模型。密钥采用 Cordis 原生方式管理：schemastery Config 带环境变量回退，通过 cordis.yml 的 `!!js process.env.MY_KEY` 注入。切勿在代码中读取自行约定的密钥文件。
+
+## 协议义务（两个实现共同验证的约定）
+
+- 在 `finish` **之前**发出 `usage`；`finish` 之后**不再发出任何内容**。稳健做法：缓冲 finish/usage 直到提供方的流结束标记，再统一 flush（可处理提供方在末尾发送仅含 usage 的分片的情况）。
+- 工具调用的 `arguments` 全程为原始 JSON 字符串；流式片段以 `argumentsDelta` 发送。如果你的提供方返回已解析的对象，请在 `block-end` 时重新 stringify。
+- 按首次出现的流顺序分配块 `index`；同一个块的每次 delta 复用该 index。
+- 错误有且仅有两条合法路径：从 `stream()` **抛出**（传输与协议故障——使用带稳定 code 的 `LlmError`），或以 `finish {kind: 'error' | 'aborted'}` 结束流（提供方带内故障）。消费方两者都处理；按故障类别选择路径并加以文档化。
+- 遵守 `options.signal`（将其传递给 fetch 或你的 SDK）。
+- 如果 `GenerateOptions` 中某个字段你的提供方无法支持（例如提供方不支持 stop sequences 时收到 `stop` 列表）：抛出 `LlmError(..., 'UNSUPPORTED_OPTION')`，而非静默丢弃。
+- 如果提供方在后续调用中需要响应 ID、签名或其他原生元数据，请将其最小无损 JSON 投影作为 `finish.replayState` 发出。重建历史时验证该状态。只有历史提供方路由和目标提供方路由当前由完全相同的适配器实例拥有时，`LlmRuntime` 才会传递该状态；由适配器决定同模型、跨模型或跨提供方恢复是否合法。状态缺失时，切勿仅根据提供方/模型名称推断原生回放。
+
+提供方特有的思考模式开关仍放在适配器的 Config 中。确切模型元数据使用一处提供方无关的能力 seam：实现 `resolveModel()`，返回提供方/模型身份以及可选的 `context` 和 `reasoning` 字段；仅当存在配置指定的默认值时才声明 `defaultEffort`；遵守解析模型时传入的可选 `AbortSignal`。推理（reasoning）强度是由适配器映射到提供方请求的有序不透明 ID。请保留适配器给出的权威可选列表，包括适配器在支持时定义的 `off`；不得暴露最终协议值的具体拼写，也不得自动调整不支持的值。ID 无需与其协议表示相同。
+
+## 实现结构
+
+让协议格式（wire format）类型、请求序列化、传输解析、分片转换和适配器类分别承担独立职责；[`llm-deepseek`](../../packages/llm/llm-deepseek/README.zh.md) 是参考布局。
+
+## 验证
+
+遵循[仓库测试策略](../testing.zh.md)，该策略负责适配器覆盖、真实提供方检查和已发布入口要求。

package/kb/extra/cookbook/extension-cookbook.md

@@ -0,0 +1,132 @@
+# Cookbook: extension plugin shapes
+
+English | [中文](extension-cookbook.zh.md)
+
+Reference patterns for harness extensions. The snippets omit imports and helper implementations and are not copy-paste-complete. For concrete authoring paths, see the [package checklist](adding-a-package.md), [first-tool tutorial](../user/develop/basic/tool.md), [tool reference](adding-a-tool.md), [LLM adapter guide](adding-an-llm-adapter.md), and [Session format version tutorial](adding-a-session-format-version.md); the [architecture](../architecture.md) owns the system and extension-point map.
+
+## A tool plugin
+
+A tool registers on `ctx.tools`. The annotated `defineTool` example (typed `execute` arguments, result construction, the `run_in_background` pattern) lives in [adding-a-tool.md](adding-a-tool.md) — that guide is the source of truth for tool definitions. Raw JSON-Schema `ToolDefinition`s are also accepted by `ctx.tools.register()` directly (that is how MCP-sourced tools arrive); `defineTool` is the typed helper for first-party tools.
+
+## A hook plugin (permission-gate example)
+
+This permission gate is one example of a hook plugin. It returns a typed decision from the `tools/pre-execute` gate to allow or deny a call; sandbox, permission, and plan-mode plugins can use this extension point. Hook plugins can intercept other extension points and are not inherently permission gates. A "native hook" is an ordinary Cordis plugin on an interception point; it needs no external protocol.
+
+```ts
+import type { Context } from '@deepseek-ai/cordis'
+import type { PreToolDecision, ToolExecution } from '@deepseek-ai/dsh-tools'
+
+declare function isAllowed(exec: ToolExecution): Promise<boolean>
+
+export const name = 'permission-gate'
+
+export function apply(ctx: Context) {
+  ctx.on('tools/pre-execute', async (exec, next): Promise<PreToolDecision> => {
+    if (!(await isAllowed(exec))) {
+      return { kind: 'deny', reason: 'Denied by policy.' }
+    }
+    return next()
+  })
+}
+```
+
+This waterfall is the reorderable policy layer. Use `ctx.tools.guard()` when an invariant needs a monotonic final denial, `tools/execute` when a plugin must wrap the dispatch lifetime (timeouts/retries/metrics; only `exec.signal` is replaceable), `tools/post-execute` for explicit result transformation, and `tools/result` for contained observation of the immutable final outcome. The [adding-a-tool guide](adding-a-tool.md#execution-policy-and-observation) gives the selection rule.
+
+## A UI plugin
+
+A UI plugin combines durable `session/event` records (Assistant settlements, turn/step boundaries, and tool activity) with transient `agent/assistant-stream` frames for live token presentation, and drives input back in via `agent.followup()` / `agent.steer()`. A browser plugin contributing a business row to the built-in Web Client instead registers a `ConversationNodeDefinition` and keyed Chat renderer; follow the [Conversation subsystem reference](../subsystems/conversation.md).
+
+```ts
+import type { Context } from '@deepseek-ai/cordis'
+import { brandString } from '@deepseek-ai/dsh-brand'
+import { createUserMessage } from '@deepseek-ai/dsh-llm'
+import type { SessionId } from '@deepseek-ai/dsh-session'
+
+declare function render(text: string): void
+declare function onUserInput(handler: (text: string) => void): void
+
+export const name = 'my-ui'
+export const inject = ['agents']
+
+export function apply(ctx: Context) {
+  ctx.on('agent/assistant-stream', ({ frame }) => {
+    if (frame.type === 'chunk' && frame.chunk.type === 'text-delta') {
+      render(frame.chunk.text)
+    }
+  })
+  onUserInput(text => ctx.agents.get(brandString<SessionId>('client-session'))?.followup(createUserMessage({
+    content: [{ type: 'text', text }],
+    source: { kind: 'user' },
+  })))
+}
+```
+
+## An external protocol driver
+
+A *protocol driver* adapts a wire peer to `ctx.agents`; it may serve a UI or an automation client. A stdio driver owns stdout, creates or resumes agents through the factory, and maps protocol requests to `followup()` or `cancel()`. A low-level prompt request returns its durable enqueue receipt; it does not acquire a result by correlating `MessageId` with `turn/end`. Publish whole-agent status separately. An automation method may wait from its receipt through the next idle and summarize that explicitly owned interval, while a UI normally keeps observing the open-ended event stream. Tear agents down with `AgentHandle.dispose()` so disposal reaches quiescence.
+
+[`packages/acp/acp`](../../packages/acp/acp) is the automation-only worked example: it exposes fresh text sessions over Agent Client Protocol JSON-RPC stdio, emits committed assistant text, and registers a one-shot machine permission answerer for agents it owns. Its [README](../../packages/acp/acp/README.md) defines the exact methods, event order, and lifecycle contract.
+
+```ts
+import type { Context } from '@deepseek-ai/cordis'
+import { expandAssistantStream } from '@deepseek-ai/dsh-llm'
+
+export const name = 'my-protocol-bridge'
+export const inject = ['agents', 'sessions', 'sessionPersistence']
+
+export function apply(ctx: Context) {
+  // Publish every committed Assistant text delta to the client.
+  ctx.on('session/event', (_session, event) => {
+    if (event.type === 'assistant/message' || event.type === 'assistant/attempt') {
+      for (const { chunk } of expandAssistantStream(event.data.stream)) {
+        if (chunk.type === 'text-delta') {
+          // sendToClient({ kind: 'message_chunk', text: chunk.text })
+        }
+      }
+    }
+  })
+  // Inbound "prompt": create/resume an agent, feed it, and return its enqueue receipt.
+  // Whole-agent status is a separate notification; no turn end belongs to this prompt.
+  // Teardown reaches quiescence via AgentHandle.dispose() (stop + await exit).
+}
+```
+
+## Runnable wirings
+
+Shipped applications contribute profile layers through `packages/bundle/*/cordis.patch.yml`, and the product `dsh` launcher owns Web, ACP, SDK, and one-shot headless execution through named profiles. Optional user-facing overlays live under `apps/cli/config/examples/`; profile integration tests live under `apps/cli/tests/profiles/`, while package-specific Loader compositions stay with their package tests.
+
+## The feature → mechanism map
+
+Every product feature maps to a listener on a documented extension point — the microkernel claim made checkable ([microkernel reference](../architecture.md)). No row modifies the loop.
+
+`system-prompt/assemble` is an expert cooperative whole-assembly transform: its returned assembly is authoritative, so listener authors own preserving active PTC mode and structured-output protocol contributions. Prefer `ctx.tools.restrict()` for tool filtering that must stay aligned across presentation, lookup, and execution.
+
+| Product feature | Plugin mechanism |
+|---|---|
+| Hook system (user + project level) | listeners on `agent/created`, `agent/pre-step`, `agent/request`, `tools/pre-execute`, `tools/post-execute`, and `agent/turn-stopping`; the waterfalls return typed decisions, while `agent/turn-stopping` may steer another step; the `dsh-hooks-claude-code` / `dsh-hooks-codex` bridges map hook config files onto these extension points |
+| `/goal` | `ctx.goals` owns durable state, `dsh-goal-round-driver` schedules same-session rounds through the public `Agent`, and separate command/tool producers expose human/model control |
+| `/loop` | on the `turn/end` session event, `followup()` the next iteration; or force-continue |
+| Dynamic workflow | `ctx.workflowEngine` + the PTC workflow engine + the `workflow` tool; structured in-process children enforce output with scoped prompt/tool registrations, a monotonic tool guard, final `tools/result` commit (including enclosing `run_code`), and the structured-output execution's monotonic `concludeTurn()` marker |
+| Queued + steering messages | core `Agent.followup()` / `Agent.steer()` |
+| Context compaction (auto + manual) | the `ctx.compaction` seam + `dsh-compaction-basic`; automatic pressure runs on serial `agent/pre-step`, canonical overflow recovery runs on `agent/request-error`, and manual callers use the same compact service ([compaction reference](../../packages/compaction/compaction/README.md)) |
+| System prompt configurability | `ctx.systemPrompt.section()` with ordering and scope-local shadowing |
+| AGENTS.md (root) | a section provider reading the file |
+| AGENTS.md (subdir, on-touch) + file-change notices | `agent.inject()` from a watcher / tool-result listener |
+| Built-in tools | `ctx.tools.register()`; schemas flow into the assembly automatically — the `dsh-tool-*` families (bash, fs, web, subagent, todo) are the shipped examples |
+| ToolSearch / progressive disclosure | replace a scoped `ctx.tools.restrict()` registration as the visible set changes; the registry keeps presentation, lookup, and execution aligned |
+| Tool deadline / retry / metrics | wrap core dispatch with `tools/execute`; a wrapper may replace `exec.signal`, delegate, and inspect the normalized result in one lexical lifetime |
+| Final tool-result metrics / audit / capture | observe immutable authoritative outcomes with `tools/result`; use `tools/post-execute` instead only when the plugin must transform the result or attach context |
+| Monotonic terminal turn policy | call `ToolExecution.concludeTurn()` from the successful terminal tool; later tool calls in the same response remain guardable, and the loop stops after the step |
+| Subprocess sandbox (landlock / sandbox-exec) | use a `ctx.sandbox` backend through `dsh-bash-sandbox`; use `tools/pre-execute` for capability-level denial |
+| Permission system / AskUserQuestion | return `ask` from `tools/pre-execute` and answer through `ctx.approval`; register a separate model-facing ask tool for ordinary user questions |
+| Plan mode | [`@deepseek-ai/dsh-plan-mode`](../../packages/plan/plan-mode/README.md) — logged `plan/mode` state, the `plan:policy` guidance section, `/plan [message]` entry, `/plan off` direct exit, and the user-reviewed `exit_plan_mode` exit; enforcement stays on the independent sandbox/approval axes |
+| Sub-agent delegation | the `ctx.subagents` provider registry (`dsh-subagent-spawn-in-process`/`dsh-subagent-fork-in-process`/`dsh-subagent-acp`/`dsh-subagent-codex`/`dsh-subagent-claude-code`/`dsh-subagent-dsh-sdk`) + `dsh-tool-subagent` exposing one configured provider to the model |
+| MCP | one plugin per server: discover tools → `ctx.tools.register()` |
+| Skills | section + tool registration; `inject()` skill content on invocation |
+| Memory | section provider + tool |
+| Scheduled tasks (cron) | a plugin registers model-callable scheduling tools; a due timer delivers through `followup(…, {source: {kind: 'schedule'}})` |
+| UI (GUI; CLI emits JSONL) | listen to `agent/assistant-stream` for live chunks and `session/event` for durable settlements, boundaries, and tool activity; input → `followup()` |
+| Web Client Chat business node | register a `ConversationNodeDefinition` and `conversation.chat.node` keyed renderer |
+| SessionTelemetryBackend / replayable trace | `session/event` → JSONL; replay = `sessions.create(id, { seed })` |
+| Model adapters | `LlmAdapter` subclass via `registerAdapter` (`dsh-llm-deepseek`, `dsh-llm-pi-ai`) |
+| Plugin hot-reload | every registration is a `ctx.effect` → vendored HMR just works |