Add in-process subagent backends: spawn (fresh) and fork (seeded)

The second PR of the subagent seam: the two in-process backends that run a
child agent on the same cordis context, reusing the agent factory's quiescent
AgentHandle teardown. Both register on ctx.subagents (PR1's named-provider
registry) and share one run driver.

- dsh-subagent-spawn: a FRESH child via ctx.agents.create — own session, the
  parent's model by default (overridable), zero inherited conversation. Also
  exports the shared in-process run driver (startInProcessRun): mint ids, stamp
  cwd/parentSession-lineage/depth, drive the one-shot (send → whenIdle), read
  the last assistant/message + turn/end reason, dispose to quiescence.
- dsh-subagent-fork: a child SEEDED with the parent's balanced completed-turn
  prefix (the log up to and including its last turn/end), so the child inherits
  context. The in-flight unbalanced turn is excluded — a raw seed would fail the
  invariants replay. Proven: a regression test goes red if the boundary seeds
  the open turn.
- Seam extension: CreateAgentOptions.seed, threaded through AgentLoop.createAgent
  → ctx.sessions.prepare({ seed }) (the primitive resume already used). This is
  the fork-lineage path the TODO(sub-agents) markers anticipated.
- Depth: a merge-extensible AgentOptions.subagentDepth (0 top-level, parent+1 for
  a child); the depthLimit capability refuses a spawn past request.maxDepth.

Tests: real-loop unit tests for both backends (mock MODEL only, real loop +
invariants), a multi-subagent test (one parent drives a fork AND a spawn child
then keeps working), and a with-key e2e (a real parent delegates via the
`subagent` tool to a real child that writes a file on disk — world-verified).
100% per-file coverage. The coding-agent demo wires the spawn backend + tool.

Snapshot coverage of nested agents is deferred to a stacked follow-up
(TODO(subagent-snapshots)): dsh-llm-replay is a single global positional cursor
that cannot route calls to a parent vs. a child on one context. Recorded in the
RFC's deferrals and a new AGENTS.md rule: designing a subsystem must design its
test infrastructure END TO END up front, verifying the snapshot/e2e harness can
express the new shape — a gap this plan hit.
This commit is contained in:
Tianyi Cui
2026-06-22 05:58:40 +08:00
parent 861791d2d8
commit 7aabd2a3df
28 changed files with 1296 additions and 13 deletions

View File

@@ -0,0 +1,23 @@
# @deepseek-ai/dsh-subagent-fork
The in-process **fork** subagent backend: a [`SubagentProvider`](../subagent/README.md) that runs each child as a child [`Agent`](../../core/agent) **seeded with a prefix of the parent's session log** — so the child inherits the parent's conversation context instead of starting fresh. Shares the run driver (`startInProcessRun`) with [`dsh-subagent-spawn`](../subagent-spawn/README.md); the only difference is the seed.
## The seed boundary (the crux)
At the moment a subagent tool's `execute` runs, the parent's CURRENT turn is open and unbalanced: the log holds the `assistant/message` carrying this spawn's tool-call and the dangling `tool/call` with no `tool/result` yet. Seeding that raw prefix would give the child an open turn that the session constructor and the dev-mode [invariants](../../support/invariants) replay **reject**.
So the fork seeds only the **balanced completed-turn prefix** — the parent's log up to and including its last `turn/end`, excluding the in-flight turn entirely (`completedTurnPrefix`). Because the live log keeps `seq === index`, the slice is contiguous-from-0 and a valid seed. A parent on its very first (not-yet-complete) turn forks an *empty* seed — i.e. effectively a fresh child.
The seam this rides on: `CreateAgentOptions.seed` (added on `dsh-agent`, threaded through `AgentLoop.createAgent``ctx.sessions.prepare({ seed })`), the same primitive `resume` uses.
## Capabilities
`{ outputSchema: false, depthLimit: true, toolFilter: false }` — identical to spawn (the depth/model/output behavior is the shared driver's).
## Config
| Key | Meaning |
|---|---|
| `providerName` | Registry name on `ctx.subagents` (default `fork`). |
See [`dsh-subagent-spawn`](../subagent-spawn/README.md) for the run lifecycle, model inheritance, and depth tracking — all shared.

View File

@@ -0,0 +1,45 @@
{
"name": "@deepseek-ai/dsh-subagent-fork",
"description": "In-process fork subagent backend: runs a child agent seeded with a prefix of the parent's log",
"version": "0.0.1",
"private": true,
"type": "module",
"main": "lib/index.js",
"types": "lib/index.d.ts",
"exports": {
".": {
"types": "./lib/index.d.ts",
"default": "./lib/index.js"
},
"./src/*": "./src/*",
"./package.json": "./package.json"
},
"files": [
"lib",
"src"
],
"license": "BSD-3-Clause",
"peerDependencies": {
"@deepseek-ai/dsh-agent": "^0.0.1",
"@deepseek-ai/dsh-session": "^0.0.1",
"@deepseek-ai/dsh-subagent": "^0.0.1",
"@deepseek-ai/dsh-subagent-spawn": "^0.0.1",
"cordis": "^4.0.0-rc.6"
},
"dependencies": {
"schemastery": "^3.18.0"
},
"devDependencies": {
"@deepseek-ai/dsh-agent": "workspace:^",
"@deepseek-ai/dsh-agent-loop": "workspace:^",
"@deepseek-ai/dsh-invariants": "workspace:^",
"@deepseek-ai/dsh-llm": "workspace:^",
"@deepseek-ai/dsh-session": "workspace:^",
"@deepseek-ai/dsh-subagent": "workspace:^",
"@deepseek-ai/dsh-subagent-spawn": "workspace:^",
"@deepseek-ai/dsh-system-prompt": "workspace:^",
"@deepseek-ai/dsh-tools": "workspace:^",
"@cordisjs/plugin-loader": "^1.0.0-rc.4",
"cordis": "^4.0.0-rc.6"
}
}

View File

@@ -0,0 +1,79 @@
/**
* The in-process FORK subagent backend: registers a {@link SubagentProvider} on
* `ctx.subagents` that runs each child as a child {@link Agent} SEEDED with a
* prefix of the parent's session log — so the child inherits the parent's
* conversation context instead of starting fresh. Shares the run driver with
* `@deepseek-ai/dsh-subagent-spawn`; the only difference is the seed.
*
* The seed boundary is the crux: at the moment a subagent tool's `execute`
* runs, the parent's CURRENT turn is open and unbalanced (it holds the
* `assistant/message` with this spawn's tool-call, plus the dangling `tool/call`
* with no `tool/result`). Seeding that raw prefix gives the child an open turn
* the session constructor and the dev-mode invariants replay REJECT. So the
* fork seeds only the **balanced completed-turn prefix**: the parent's log up
* to and including its last `turn/end`, excluding the in-flight turn entirely.
*
* Plugin export shape: named `name`/`inject`/`Config`/`apply`, NO default.
*
* @module @deepseek-ai/dsh-subagent-fork
*/
import type { Context } from 'cordis'
import z from 'schemastery'
import type { SessionEvent } from '@deepseek-ai/dsh-session'
import type { Agent } from '@deepseek-ai/dsh-agent'
import type { SubagentCapabilities, SubagentProvider, SubagentStartRequest } from '@deepseek-ai/dsh-subagent'
import { startInProcessRun } from '@deepseek-ai/dsh-subagent-spawn'
export const name = 'subagent-fork'
export const inject = ['subagents', 'agents']
/** Config: the registry name to register the provider under. */
export interface Config {
/** Provider name on `ctx.subagents` (default `fork`). */
providerName: string
}
export const Config: z<Config> = z.object({
providerName: z.string().default('fork'),
})
/**
* The balanced completed-turn prefix of `parent`'s log: every event up to and
* including the last `turn/end`. Empty if the parent has never completed a turn
* (the in-flight turn is excluded, so a parent on its very first turn forks an
* empty — i.e. fresh — child). The result is contiguous from seq 0 (the live
* log keeps `seq === index`), so it is a valid session seed; the in-flight,
* unbalanced turn is dropped so the invariants replay accepts it.
*/
export function completedTurnPrefix(parent: Agent): SessionEvent[] {
const events = parent.session.events
const lastEnd = events.findLast(e => e.type === 'turn/end')
if (lastEnd === undefined) return []
// seq === array index (the append contract), so slice up to and including it.
return events.slice(0, lastEnd.seq + 1)
}
/**
* The fork provider. Supports `depthLimit`; NOT `outputSchema`/`toolFilter` this
* cut (the service rejects a request needing either before `start` runs).
*/
class ForkProvider implements SubagentProvider {
readonly capabilities: SubagentCapabilities = { outputSchema: false, depthLimit: true, toolFilter: false }
constructor(readonly name: string, private readonly ctx: Context) {}
start(request: SubagentStartRequest) {
const seed = completedTurnPrefix(request.parent)
return startInProcessRun(this.ctx, request, {
providerName: this.name,
// Only pass a seed when there's a completed turn to inherit; an empty seed
// is equivalent to a fresh child, so omit it to keep the session unseeded.
...seed.length > 0 ? { seed } : {},
})
}
}
export function apply(ctx: Context, config: Config): void {
ctx.subagents.registerProvider(new ForkProvider(config.providerName, ctx))
}

View File

@@ -0,0 +1,99 @@
import { describe, expect, it } from 'vitest'
import { Context } from 'cordis'
import LlmService from '@deepseek-ai/dsh-llm'
import SessionStore from '@deepseek-ai/dsh-session'
import SystemPrompt from '@deepseek-ai/dsh-system-prompt'
import ToolRegistry from '@deepseek-ai/dsh-tools'
import AgentRegistry, { AgentId } from '@deepseek-ai/dsh-agent'
import AgentLoop from '@deepseek-ai/dsh-agent-loop'
import * as Invariants from '@deepseek-ai/dsh-invariants'
import SubagentService from '@deepseek-ai/dsh-subagent'
import * as Spawn from '@deepseek-ai/dsh-subagent-spawn'
import { MockAdapter, textResponse } from '../../../core/agent-loop/tests/mock-adapter.ts'
import * as fork from '../src/index.ts'
type Script = ConstructorParameters<typeof MockAdapter>[0]
/**
* The two in-process backends coexist on one context: the SAME parent agent
* delegates to a `spawn` child (fresh) and a `fork` child (seeded with its log),
* and keeps working itself. This is the multi-provider coexistence the seam
* exists for — the named registry lets one runtime hold both transports.
*/
async function setup(script: Script) {
const ctx = new Context()
await ctx.plugin(LlmService)
await ctx.plugin(SessionStore)
await ctx.plugin(SystemPrompt)
await ctx.plugin(ToolRegistry)
await ctx.plugin(AgentRegistry)
await ctx.plugin(Invariants)
await ctx.plugin(AgentLoop, { agents: [] })
await ctx.plugin(SubagentService)
await ctx.plugin(Spawn, { providerName: 'spawn' })
await ctx.plugin(fork, { providerName: 'fork' })
ctx.llm.registerAdapter(['mock'], new MockAdapter(script))
const parent = ctx.agentLoop.create(AgentId('parent'), { model: 'mock' })
return { ctx, parent }
}
function text(blocks: { type: string; text?: string }[]): string {
return blocks.filter(b => b.type === 'text').map(b => b.text).join('')
}
describe('multi-subagent coexistence (spawn + fork on one context)', () => {
it('both providers register and coexist', async () => {
const { ctx } = await setup([])
expect(ctx.subagents.list().sort()).toEqual(['fork', 'spawn'])
})
it('the same parent drives a spawn child AND a fork child, then keeps working', async () => {
// Script order: parent turn 1, spawn child, fork child, parent turn 2.
const { ctx, parent } = await setup([
textResponse('parent turn one'),
textResponse('spawn child reply'),
textResponse('fork child reply'),
textResponse('parent turn two'),
])
// Parent does one real turn first, so the fork has a completed turn to seed.
parent.send([{ type: 'text', text: 'parent q1' }])
await parent.whenIdle()
const parentPrefixLen = parent.session.events.length
// Delegate to a fresh spawn child.
const spawnRun = ctx.subagents.start('spawn', { prompt: [{ type: 'text', text: 'spawn task' }], parent })
const spawnResult = await spawnRun.result
expect(spawnResult.stopReason).toBe('completed')
expect(text(spawnResult.output)).toBe('spawn child reply')
// Delegate to a fork child (seeded with the parent's turn-1 prefix).
const forkRun = ctx.subagents.start('fork', { prompt: [{ type: 'text', text: 'fork task' }], parent })
const forkResult = await forkRun.result
expect(forkResult.stopReason).toBe('completed')
expect(text(forkResult.output)).toBe('fork child reply')
// The two children are distinct sessions, both lineage-stamped to the parent.
const spawnChild = ctx.agents.get(spawnRun.id)!
const forkChild = ctx.agents.get(forkRun.id)!
expect(spawnChild.session.header.id).not.toBe(forkChild.session.header.id)
expect(spawnChild.session.header.parentSession).toBe(parent.session.header.id)
expect(forkChild.session.header.parentSession).toBe(parent.session.header.id)
// The fork child inherited the parent's prefix; the spawn child did not.
expect(forkChild.session.events.slice(0, parentPrefixLen).some(e => e.type === 'user/message')).toBe(true)
await spawnRun.dispose()
await forkRun.dispose()
// The parent is unaffected and keeps working after both delegations.
parent.send([{ type: 'text', text: 'parent q2' }])
await parent.whenIdle()
const lastParentMessage = parent.session.events.findLast(e => e.type === 'assistant/message')
expect(lastParentMessage?.type === 'assistant/message' && text(lastParentMessage.data.content)).toBe('parent turn two')
// The parent's OWN log never recorded the children's internal steps — its
// only subagent-related entries would be tool/call+tool/result IF it had
// used the tool, but here we called the service directly, so the parent log
// is purely its own two turns.
expect(parent.session.events.filter(e => e.type === 'turn/end')).toHaveLength(2)
})
})

View File

@@ -0,0 +1,161 @@
import { describe, expect, it } from 'vitest'
import { Context } from 'cordis'
import Loader from '@cordisjs/plugin-loader'
import LlmService from '@deepseek-ai/dsh-llm'
import SessionStore from '@deepseek-ai/dsh-session'
import SystemPrompt from '@deepseek-ai/dsh-system-prompt'
import ToolRegistry from '@deepseek-ai/dsh-tools'
import AgentRegistry, { AgentId } from '@deepseek-ai/dsh-agent'
import AgentLoop from '@deepseek-ai/dsh-agent-loop'
import * as Invariants from '@deepseek-ai/dsh-invariants'
import SubagentService from '@deepseek-ai/dsh-subagent'
import { MockAdapter, textResponse } from '../../../core/agent-loop/tests/mock-adapter.ts'
import * as fork from '../src/index.ts'
import { completedTurnPrefix } from '../src/index.ts'
type Script = ConstructorParameters<typeof MockAdapter>[0]
/**
* Drives the REAL fork backend with a real loop + scripted mock MODEL + the
* real dsh-invariants plugin. The invariants plugin re-replays a seeded child
* log on `session/created` (its freeze-check), so a malformed (unbalanced) fork
* seed makes these tests THROW — that is the regression guard for the
* completed-turn-prefix boundary.
*/
async function setup(script: Script) {
const ctx = new Context()
await ctx.plugin(LlmService)
await ctx.plugin(SessionStore)
await ctx.plugin(SystemPrompt)
await ctx.plugin(ToolRegistry)
await ctx.plugin(AgentRegistry)
await ctx.plugin(Invariants)
await ctx.plugin(AgentLoop, { agents: [] })
await ctx.plugin(SubagentService)
await ctx.plugin(fork, { providerName: 'fork' })
ctx.llm.registerAdapter(['mock'], new MockAdapter(script))
const parent = ctx.agentLoop.create(AgentId('parent'), { model: 'mock' })
return { ctx, parent }
}
function text(blocks: { type: string; text?: string }[]): string {
return blocks.filter(b => b.type === 'text').map(b => b.text).join('')
}
describe('completedTurnPrefix', () => {
it('returns an empty prefix for a parent that has never completed a turn', async () => {
const { parent } = await setup([])
expect(completedTurnPrefix(parent)).toEqual([])
})
it('returns the balanced prefix up to and including the last turn/end', async () => {
const { parent } = await setup([textResponse('first'), textResponse('second')])
parent.send([{ type: 'text', text: 'q1' }])
await parent.whenIdle()
parent.send([{ type: 'text', text: 'q2' }])
await parent.whenIdle()
const prefix = completedTurnPrefix(parent)
// Ends exactly at the last turn/end; seq is contiguous from 0.
expect(prefix.at(-1)?.type).toBe('turn/end')
expect(prefix.map(e => e.seq)).toEqual(prefix.map((_, i) => i))
// Both completed turns are present.
expect(prefix.filter(e => e.type === 'turn/end')).toHaveLength(2)
})
})
describe('dsh-subagent-fork', () => {
it('forks an UNSEEDED (fresh) child when the parent has no completed turn', async () => {
// The parent has never completed a turn → empty prefix → the provider omits
// the seed → the child runs fresh. Exercises the `seed.length > 0` false arm.
const { ctx, parent } = await setup([textResponse('fresh child')])
expect(completedTurnPrefix(parent)).toEqual([])
const run = ctx.subagents.start('fork', { prompt: [{ type: 'text', text: 'child q' }], parent })
const result = await run.result
expect(result.stopReason).toBe('completed')
expect(text(result.output)).toBe('fresh child')
const child = ctx.agents.get(run.id)!
// Only the child's own turn — no seeded parent turns.
expect(child.session.events.filter(e => e.type === 'turn/end')).toHaveLength(1)
await run.dispose()
})
it('seeds the child with the parent\'s completed-turn prefix (child inherits context)', async () => {
// Parent runs one turn, then we fork. The child's seeded log should contain
// the parent's first turn, and the child should run its own new turn on top.
const { ctx, parent } = await setup([textResponse('parent answer'), textResponse('child answer')])
parent.send([{ type: 'text', text: 'parent question' }])
await parent.whenIdle()
const parentPrefixLen = parent.session.events.length
const run = ctx.subagents.start('fork', { prompt: [{ type: 'text', text: 'child question' }], parent })
const result = await run.result
expect(result.stopReason).toBe('completed')
expect(text(result.output)).toBe('child answer')
const child = ctx.agents.get(run.id)!
// The child's log STARTS with the parent's prefix (seeded), then its own turn.
expect(child.session.events.length).toBeGreaterThan(parentPrefixLen)
// The seeded prefix carried the parent's user message.
const seededUser = child.session.events.slice(0, parentPrefixLen).find(e => e.type === 'user/message')
expect(seededUser).toBeDefined()
// Lineage stamped.
expect(child.session.header.parentSession).toBe(parent.session.header.id)
await run.dispose()
})
it('produces an invariant-CLEAN seed: forking mid-turn excludes the open turn', async () => {
// Drive the parent so it has ONE completed turn, then start a SECOND turn
// that is still open (a hanging model call), and fork while it's in flight.
// The fork must seed only the completed first turn — an unbalanced seed
// would make the invariants replay throw inside ctx.subagents.start.
const { ctx, parent } = await setup([textResponse('done'), 'hang', textResponse('child')])
parent.send([{ type: 'text', text: 'q1' }])
await parent.whenIdle()
// Start a second turn that hangs (open turn/start + open step, never ends).
parent.send([{ type: 'text', text: 'q2' }])
await new Promise(r => setTimeout(r, 20)) // let the hanging turn open
// Forking now must NOT throw (the open second turn is excluded from the seed).
const run = ctx.subagents.start('fork', { prompt: [{ type: 'text', text: 'child q' }], parent })
const result = await run.result
expect(result.stopReason).toBe('completed')
expect(text(result.output)).toBe('child')
const child = ctx.agents.get(run.id)!
// The child's seed has exactly the ONE completed parent turn (the open one excluded).
const seedTurnEnds = child.session.events.filter(e => e.type === 'turn/end')
// 1 from the seeded parent turn + 1 from the child's own completed turn.
expect(seedTurnEnds.length).toBe(2)
parent.cancel()
await run.dispose()
})
it('advertises depthLimit but not outputSchema/toolFilter', async () => {
const { ctx } = await setup([])
expect(ctx.subagents.getProvider('fork')!.capabilities).toEqual({ outputSchema: false, depthLimit: true, toolFilter: false })
})
it('unregisters the provider when its fiber is disposed (HMR safety)', async () => {
const ctx = new Context()
await ctx.plugin(SubagentService)
await ctx.plugin(AgentRegistry)
const fiber = await ctx.plugin(fork, { providerName: 'fork' })
expect(ctx.subagents.list()).toEqual(['fork'])
await fiber.dispose()
expect(ctx.subagents.list()).toEqual([])
})
it('has the namespace-plugin export shape (no stray default)', () => {
expect('default' in fork).toBe(false)
expect(fork.name).toBe('subagent-fork')
expect(fork.inject).toEqual(['subagents', 'agents'])
const loader = Object.create(Loader.prototype) as Loader
const unwrapped = loader.unwrapExports(fork) as Record<string, unknown>
expect(unwrapped).toBe(fork)
expect(unwrapped.name).toBe('subagent-fork')
expect(unwrapped.inject).toEqual(['subagents', 'agents'])
expect(typeof unwrapped.apply).toBe('function')
})
})

View File

@@ -0,0 +1,33 @@
{
"extends": "../../../tsconfig.base.json",
"compilerOptions": {
"rootDir": "src",
"outDir": "lib"
},
"include": [
"src"
],
"references": [
{
"path": "../../../vendor/cosmokit"
},
{
"path": "../../../vendor/cordis"
},
{
"path": "../../../vendor/schemastery"
},
{
"path": "../../core/agent"
},
{
"path": "../../core/session"
},
{
"path": "../subagent"
},
{
"path": "../subagent-spawn"
}
]
}