fix: address codex review round 1

- Strict steer now rejects the two windows where an acknowledged
  message would be silently dropped: the closed-turn durability-flush
  window (status still running, loop strands drained steering) and a
  committed structured capture (terminal turn-stop discards late
  steering). Seam JSDoc, catalog doc, README, and the Agent Note
  bilingual pair state the tightened contract; new keyless tests pin
  both rejections.
- Continuable background delegation now fails loud when the advertised
  send_message tool is not registered, instead of starting a durable
  child the model cannot continue. The acp-agent example already loads
  the control tool; the tool-catalog boot recipe is unaffected because
  capability wording is harvested at mount.
This commit is contained in:
Dudu-0223
2026-07-23 17:53:20 +08:00
committed by imccyu
parent 3be7ca8664
commit 71570d7bec
11 changed files with 126 additions and 15 deletions

View File

@@ -30,7 +30,7 @@ The required request signal covers both startup and the live run. Before publica
After fulfillment, the caller owns the run. Provider-plugin unload does not revoke it. `dispose()` removes the live abort listener, records cancellation, and delegates to the returned `AgentHandle.dispose()`, whose memoized quiescence transaction stops the loop, removes the agent and session, and unwinds scoped registrations. Cancellation owns every non-completed in-flight outcome and reports `aborted`; an already-completed turn remains completed.
Runs expose the strict `steer` capability: a synchronous `AgentStatus.running` check and `Agent.steer()` call share one frame, so delivery joins the observed turn or throws. The Agent-level idle fallback (queue and start a new turn) is deliberately not reachable through the run — that would start an untracked turn after the run's result was read.
Runs expose the strict `steer` capability: the synchronous checks and the `Agent.steer()` call share one frame, so delivery joins the observed turn or throws. Delivery requires `AgentStatus.running`, an open turn in the child log (status stays `running` through a closed turn's durability flush, where the loop would strand the message), and no committed structured capture (whose terminal stop makes the loop discard late steering). The Agent-level idle fallback (queue and start a new turn) is deliberately not reachable through the run — that would start an untracked turn after the run's result was read.
## Spawn and fork inputs

View File

@@ -259,13 +259,30 @@ function driveTurn(
return handle.dispose()
},
steer(content: ContentBlock[]): void {
// Strict live delivery: the synchronous running check and Agent.steer()
// Strict live delivery: the synchronous checks and the Agent.steer()
// call share one frame, so delivery joins the observed turn or throws.
// Agent.steer()'s own idle fallback would instead QUEUE the message and
// start a new, untracked turn after this run's result was read.
if (child.status !== 'running') {
throw new Error(`subagent child "${childId}" is not running; the message was not delivered`)
}
// The status stays `running` through the closed turn's durability flush,
// and the loop DISCARDS terminal-stopped steering drained after turn
// close instead of recording it. Requiring an open turn keeps
// acknowledged delivery honest.
const lastBoundary = child.session.events.findLast(
event => event.type === 'turn/start' || event.type === 'turn/end',
)
if (lastBoundary?.type !== 'turn/start') {
throw new Error(`subagent child "${childId}" turn has already closed; the message was not delivered`)
}
// A committed structured capture makes the pending `agent/turn-stop`
// checkpoint terminal, and the loop then discards late steering. The
// capture is synchronously observable, so reject rather than
// acknowledge a message the run is about to drop.
if (structured?.captured() !== undefined) {
throw new Error(`subagent child "${childId}" already reported its structured result; the message was not delivered`)
}
child.steer(createUserMessage({ content, source: { kind: 'user' } }))
},
}

View File

@@ -120,6 +120,33 @@ describe('in-process structured output', () => {
await run.dispose()
})
it('strict steer rejects delivery once the structured result is captured', async () => {
// Hold the capture's tool result open so the child is observably running
// with a committed capture: the pending agent/turn-stop checkpoint is
// terminal, and the loop would DISCARD a steering message, so an
// acknowledged delivery here would be a lie.
let releaseResult: (() => void) | undefined
const { ctx, parent } = await setup([
toolCallResponse('c1', STRUCTURED_OUTPUT_TOOL, { answer: 7 }),
])
ctx.on('agent/post-step', (agent) => {
if (agent.session.header.parentSession === undefined || releaseResult !== undefined) return
return new Promise<void>((resolve) => { releaseResult = resolve })
})
const run = await ctx.subagents.start('spawn', structuredRequest(parent))
await new Promise<void>((resolve) => {
const timer = setInterval(() => {
if (releaseResult !== undefined) { clearInterval(timer); resolve() }
}, 5)
})
expect(() => { run.steer!([{ type: 'text', text: 'one more thing' }]) })
.toThrow(/already reported its structured result; the message was not delivered/)
releaseResult!()
const result = await run.result
expect(result.structured).toEqual({ answer: 7 })
await run.dispose()
})
it('denies tool calls that FOLLOW the capture in the same response — terminal means terminal', async () => {
// One model response carrying structured_output FIRST and a side-effecting
// call after it: the continuation veto only fires at step end, so without

View File

@@ -245,4 +245,45 @@ describe('startInProcessRun', () => {
expect(ctx.agents.list()).toHaveLength(beforeAgents)
expect(ctx.sessions.list()).toHaveLength(beforeSessions)
})
it('strict steer rejects a settled child instead of queueing an untracked turn', async () => {
const { ctx, parent } = await setup([textResponse('done')])
const run = await startInProcessRun(request(parent), {})
await run.result
// The child is idle after its turn: Agent.steer() would silently QUEUE.
expect(() => { run.steer!([{ type: 'text', text: 'late' }]) })
.toThrow(/not running; the message was not delivered/)
const child = ctx.agents.get(run.id)!
expect(child.session.events.some(event => event.type === 'steering/message')).toBe(false)
await run.dispose()
})
it('strict steer rejects the closed-turn flush window where the loop discards steering', async () => {
// Hold the turn-end durability flush open: the turn has closed in the log
// and status is still `running`, exactly the window where the loop would
// discard a drained steering message instead of recording it.
const { ctx, parent } = await setup([textResponse('quick')])
let releaseFlush: (() => void) | undefined
ctx.on('session/flush', (session) => {
if (session.header.parentSession === undefined || releaseFlush !== undefined) return
const lastEnd = session.events.findLast(event => event.type === 'turn/end')
if (lastEnd === undefined) return
return new Promise<void>((resolve) => { releaseFlush = resolve })
})
const run = await startInProcessRun(request(parent), {})
const child = ctx.agents.get(run.id)!
// Wait until the child's turn has closed while the flush keeps it running.
await new Promise<void>((resolve) => {
const timer = setInterval(() => {
if (releaseFlush !== undefined) { clearInterval(timer); resolve() }
}, 5)
})
expect(child.status).toBe('running')
expect(() => { run.steer!([{ type: 'text', text: 'into the void' }]) })
.toThrow(/turn has already closed; the message was not delivered/)
releaseFlush!()
await run.result
expect(child.session.events.some(event => event.type === 'steering/message')).toBe(false)
await run.dispose()
})
})

View File

@@ -219,10 +219,11 @@ export interface SubagentRun {
/**
* OPTIONAL (strict live-steering capability): deliver additional content to
* the actively running child turn. STRICT means delivery joins the observed
* turn or fails — the implementation must synchronously require the child to
* be running with no asynchronous boundary before delivery, and must not
* fall back to a queue path that could start a new, untracked turn after
* this run has settled. Throws when the child is not running. A run
* turn or fails — the implementation must synchronously verify, with no
* asynchronous boundary before delivery, that the child is running and its
* turn can still record the message, and must not fall back to a queue path
* that could start a new, untracked turn or silently drop the message after
* this run has settled. Throws when delivery cannot join the turn. A run
* represents one disposable activation, so it has no cold-resume operation;
* resuming a settled child goes through {@link SubagentProvider.resume}.
*/

View File

@@ -285,6 +285,13 @@ export function apply(ctx: Context, config: Config): void {
if (control === undefined) {
throw new Error('continuable background subagents unavailable: load @deepseek-ai/dsh-subagent-control and @deepseek-ai/dsh-tool-tasks')
}
// The schema above tells the model to follow up with
// `send_message`; starting a durable child the model cannot
// continue would make that advertisement false. Sibling load order
// is undetermined at mount, so the check lives at the operation.
if (ctx.tools.get('send_message') === undefined) {
throw new Error('continuable background subagents unavailable: load @deepseek-ai/dsh-tool-subagent-control (the advertised send_message tool is not registered)')
}
// The control service owns the durable child id, descriptor
// snapshot, Task registration, and settle-then-dispose ordering; a
// synchronous validation failure rejects the call with no Task.

View File

@@ -17,6 +17,7 @@ import type { SubagentStartRequest } from '@deepseek-ai/dsh-subagent'
import LocalTaskService from '@deepseek-ai/dsh-tasks-local'
import SubagentControlService from '@deepseek-ai/dsh-subagent-control'
import * as SubagentSpawn from '@deepseek-ai/dsh-subagent-spawn'
import * as ToolSubagentControl from '@deepseek-ai/dsh-tool-subagent-control'
import * as ToolTasks from '@deepseek-ai/dsh-tool-tasks'
import { MockAdapter, textResponse } from '../../../core/agent-loop/tests/mock-adapter.ts'
import * as mock from './scripted-provider.ts'
@@ -825,7 +826,7 @@ describe('dsh-tool-subagent continuable background mode', () => {
})
/** Boot the real continuable stack: loop, persistence, spawn, tasks, control. */
async function continuableSetup() {
async function continuableSetup(options: { controlTool?: boolean } = {}) {
const ctx = new Context()
await mountAgentLoopTestDependencies(ctx)
const root = mkdtempSync(path.join(tmpdir(), 'dsh-tool-subagent-continuable-'))
@@ -837,6 +838,7 @@ describe('dsh-tool-subagent continuable background mode', () => {
await ctx.plugin(LocalTaskService)
await ctx.plugin(ToolTasks, {})
await ctx.plugin(SubagentControlService)
if (options.controlTool !== false) await ctx.plugin(ToolSubagentControl)
await ctx.plugin(tool, { provider: 'spawn' })
ctx.llm.registerAdapter(['mock'], new MockAdapter([
textResponse('continuable answer'),
@@ -886,6 +888,21 @@ describe('dsh-tool-subagent continuable background mode', () => {
expect(result.isError).toBe(true)
expect(text(result)).toContain('load @deepseek-ai/dsh-subagent-control')
})
it('fails loud when the advertised send_message tool is not registered', async () => {
// The schema tells the model to follow up with send_message; starting a
// durable child the model cannot continue would make that false.
const { ctx, parent } = await continuableSetup({ controlTool: false })
const result = await callSubagent(
ctx,
{ description: 'd', prompt: 'p', run_in_background: true },
{ agent: parent },
)
expect(result.isError).toBe(true)
expect(text(result)).toContain('load @deepseek-ai/dsh-tool-subagent-control')
// Nothing was started: no Task exists for the parent.
expect(ctx.tasks.list(parent)).toEqual([])
})
})
describe('background preflight failure (no orphaned child, by construction)', () => {