mirror of
https://github.com/deepseek-ai/deepseek-harness
synced 2026-08-15 21:04:50 +00:00
133 lines
8.5 KiB
Markdown
133 lines
8.5 KiB
Markdown
# 工作流
|
||
|
||
[English](workflow.md) | 中文
|
||
|
||
工作流 seam：一个 agent（智能体）运行由模型编写的编排脚本（SCRIPT），扇出 subagent。与 [subagent](subagent.md) 一样，它是**一项可选能力**，不属于 agent loop（智能体循环）主干，因此其词汇定义在此处而非 [core.md](core.md)。与 subagent 注册表不同，它采用 bash 形态：每个上下文只有一个引擎实现提供 `ctx.workflows`；没有命名提供方注册表（第二个引擎是插件替换，而非共存）。
|
||
|
||
接口：[dsh-workflow](../../packages/workflow/workflow)（`ctx.workflows` + 下文词汇）。实现是 [dsh-workflow-workerthread](../../packages/workflow/workflow-workerthread)（一个 `node:worker_threads` 引擎——每个 run 一个 worker，脚本的 vm 上下文位于其中）；面向模型的消费方是 [dsh-tool-workflow](../../packages/workflow/tool-workflow)。提案与设计理由见 [dynamic-workflows Agent Note（agent 决策记录）](../../.agents/notes/implemented/feature/2026-07-05-dynamic-workflows.md)。
|
||
|
||
源码：[`packages/workflow/workflow/src/types.ts`](../../packages/workflow/workflow/src/types.ts)
|
||
|
||
## 启动请求
|
||
|
||
调用方启动 run 时提出的请求。普通工作流工具根据模型的 `{ script, meta, args }` 调用与发起调用的 agent 构建它；专用消费方还可以为该 run 选择一个引擎级 `subagentProvider` 并调低 `maxTotalAgents`，但脚本无法观察或替换这两项策略。`meta` 与 `args` 是普通 JSON 数据（引擎会对 `meta` 做形状校验，并在任何内容运行前大声拒绝——绝不会通过求值脚本文本来获取它）。`parent` 是必填字段——脚本生成的每个子 agent 都归属于它（cwd、谱系与深度通过 [subagent seam](subagent.md) 流转）。
|
||
|
||
```ts type-equiv
|
||
/**
|
||
* What a caller asks for when starting a workflow run. `meta` and `args` are
|
||
* plain JSON DATA by the seam contract (the tool builds both from the model's
|
||
* schema-validated call; the engine validates `meta`'s shape and rejects loud
|
||
* before anything runs) — an engine never evaluates script text to obtain
|
||
* them. `parent` is REQUIRED — every `agent()` the script spawns is
|
||
* attributed to it (cwd, lineage, depth flow through the subagent seam).
|
||
*/
|
||
interface WorkflowStartRequest {
|
||
/** The plain-JS script body (top-level await allowed; ends with `return <json-value>`). */
|
||
script: string
|
||
/** The workflow's identity block, as plain JSON data (shape-validated by the engine). */
|
||
meta: WorkflowMeta
|
||
/** Optional input exposed verbatim to the script as the `args` global. */
|
||
args?: unknown
|
||
/**
|
||
* Optional engine-wide child-provider override for this run. The workflow
|
||
* script cannot observe or replace it; omission uses the engine's configured
|
||
* provider.
|
||
*/
|
||
subagentProvider?: string
|
||
/**
|
||
* Optional per-run total-child ceiling. Implementations reject values above
|
||
* their deployment ceiling before publishing the run.
|
||
*/
|
||
maxTotalAgents?: number
|
||
/** The agent on whose behalf the run executes (parent of every child). */
|
||
parent: Agent
|
||
/** Cancels the run when aborted (the tool's `exec.signal`). */
|
||
signal?: AbortSignal
|
||
}
|
||
```
|
||
|
||
## 工作流的身份标识：`WorkflowMeta`
|
||
|
||
作为数据附在启动请求上的身份块（工具的 `meta` 参数；字段词汇与 Claude Code 动态工作流的 meta 块一致）。`phases` 仅用于进度展示：`phase()` 调用与标题匹配，供观察者使用；不暗示任何执行结构。
|
||
|
||
```ts type-equiv
|
||
/**
|
||
* The script's identity block, provided as plain JSON data alongside the
|
||
* script body (the model-facing tool carries it as its `meta` parameter) and
|
||
* validated by the engine before the body runs. `name`/`description` are
|
||
* required; the rest is optional annotation. The field vocabulary matches the
|
||
* Claude Code dynamic-workflows meta block.
|
||
*/
|
||
interface WorkflowMeta {
|
||
/** Short kebab-case workflow name (display + persistence key). */
|
||
name: string
|
||
/** One-line description of what the workflow does. */
|
||
description: string
|
||
/** Optional guidance on when this workflow applies (shown in listings). */
|
||
whenToUse?: string
|
||
/** Optional phase declarations matched by `phase()` calls. */
|
||
phases?: WorkflowPhase[]
|
||
}
|
||
```
|
||
|
||
## 终态结果：`WorkflowResult`
|
||
|
||
一次运行的结果，由 `WorkflowRun.result` resolve。`value` 是脚本的物化返回值——纯宿主域 JSON 数据（脚本无返回值时为 `null`）——仅在 `completed` 时有意义。`stopReason` 是封闭联合类型（引擎所有；消费方可穷举）：`completed` | `cancelled` | `error`。非 `completed` 的原因在 `error` 中携带失败信息，消费方将其映射为 `isError` 工具结果，而非把部分输出当作成功上报。
|
||
|
||
```ts type-equiv
|
||
/**
|
||
* The outcome of one run, resolved by {@link WorkflowRun.result}. `value` is
|
||
* the script's materialized return value (plain host-realm JSON data; `null`
|
||
* when the script returned `undefined`) — meaningful only for `completed`.
|
||
* A non-`completed` reason carries the failure in `error`; the consumer maps
|
||
* it to an `isError` tool result rather than reporting partial output.
|
||
*/
|
||
interface WorkflowResult {
|
||
/** The script's return value (host JSON data; `null` for no return). */
|
||
value: unknown
|
||
/** Why the run settled. */
|
||
stopReason: WorkflowStopReason
|
||
/** The failure message (present iff `stopReason` is not `completed`). */
|
||
error?: string
|
||
/**
|
||
* How many `agent()` calls the run accepted over its whole lifetime. On a
|
||
* graceful settlement this is the script-side count (calls still queued for
|
||
* a concurrency slot included); on a termination path (grace force-settle,
|
||
* worker death) it degrades to the host-observed count — calls queued
|
||
* inside a terminated script are unknowable then.
|
||
*/
|
||
agentsStarted: number
|
||
}
|
||
```
|
||
|
||
## 活跃运行：`WorkflowRun`
|
||
|
||
脚本执行期间消费方持有的句柄。消费方 await `result`，可中途 `cancel`，且必须在每条路径上 `dispose`（资源释放）。`result` 不会 reject：脚本失败以 `stopReason: 'error'` resolve；一旦运行被取消，即使脚本本身永不 settle，它也会在引擎的有界宽限期内 settle（引擎强制以 `cancelled` settle；worker-thread 引擎随后终止脚本的 worker），因此消费方 await `result` 不会在取消后卡死。`dispose()` = cancel + 有界 settle + 等待子 agent 停稳；它不会因脚本卡死而挂起。
|
||
|
||
```ts type-equiv
|
||
/**
|
||
* Holder-owned live workflow. `result` never rejects and settles within the
|
||
* engine's cancellation grace; failures resolve through `stopReason`. Consumers
|
||
* may cancel and must call idempotent `dispose()` on every path to await bounded
|
||
* script settlement and child quiescence.
|
||
*/
|
||
interface WorkflowRun {
|
||
readonly id: WorkflowRunId
|
||
/** The validated meta block (available before the body runs). */
|
||
readonly meta: WorkflowMeta
|
||
readonly result: Promise<WorkflowResult>
|
||
/** Cancel the run: children abort, pending hooks reject, the script dies at its next await (or is force-settled at the grace). */
|
||
cancel(reason?: string): void
|
||
/** Cancel + bounded-grace settle; safe to call on every path (idempotent). */
|
||
dispose(): Promise<void>
|
||
}
|
||
```
|
||
|
||
## 失败纪律：`WorkflowError.fatal`
|
||
|
||
脚本内部的钩子误用：错误参数、未知或延迟的 `agent()` 选项、超出[结构化输出子集](../../packages/core/tools/README.md)的 schema、触发的上限、seam 启动失败、取消，都会抛出 `fatal: true` 的 `WorkflowError`。`parallel()`/`pipeline()` 组合器对 fatal 错误直接重新抛出，而非将该项映射为 `null`：一个拼写错误的选项必须让脚本大声失败，绝不能消融为看似普通子 agent 失败的结果。逐项的 `null` 保留给子运行失败（非 `completed` 的 stop reason）和阶段内的普通脚本错误。
|
||
|
||
## 事件
|
||
|
||
`workflow/*` 事件（`workflow/start`、`workflow/phase`、`workflow/log`、`workflow/agent-start`、`workflow/agent-end`、`workflow/end`，见[事件目录](../cordis-catalog/events.md)）是**仅供观察**的 emit，携带数据快照：每个 payload 以 `WorkflowRunInfo`（id + meta）开头，而非活跃的 `WorkflowRun`，因此订阅者无法获得 `cancel`/`dispose`；`workflow/end` 刻意省略 result value（观察结果的监听器不得收到调用方 result 的可变别名）。每次 emit 对每个监听器隔离：抛出异常的订阅者被记录日志但不传播，不会饿死在它之后注册的监听器；每个监听器收到自己的 payload 克隆，因此修改它既不会损坏引擎也不会影响其他监听器。这种隔离方式与 `subagent/start`/`subagent/end` 一致。
|