Merge remote-tracking branch 'origin/codex/enforce-tool-cancellation' into worktree/explicit-turn-signal

# Conflicts:
#	docs/architecture.md
#	docs/cordis-catalog/events.md
#	docs/core-data-structures/core.md
#	docs/event-producer-consumer.md
#	packages/core/agent-loop/README.md
#	packages/core/agent-loop/src/agent.ts
#	packages/core/agent/README.md
#	packages/core/agent/src/types.ts
#	packages/ui/acp/src/index.ts
#	packages/ui/tui/src/index.ts
#	packages/ui/tui/tests/harness.ts
This commit is contained in:
Tianyi Cui
2026-07-21 12:48:46 +08:00
267 changed files with 13024 additions and 278 deletions

View File

@@ -56,7 +56,7 @@ The driver owns one agent for its lifetime and runs inside `ctx.agents.withIniti
Every provider call that reaches a successful finish appends exactly one `assistant/message` completion anchor, including content-less calls and `max-tokens` finishes. A successful `agent/step-result` stores its transformed content; a rejected result records empty content before the original failure continues. The anchor retains exact chunk provenance (`[]` for a stream with no chunks) and usage when available, while empty content stays out of derived message history.
Plugin failure ends the current turn, not the loop. Only final adapter dispatch/iteration failures and terminal in-band error or aborted finishes enter `agent/request-error`; middleware, result processing, tools, and `agent/post-step` remain ordinary turn failures. Recovery receives the exact live error, immutable provider facts, and immutable prior failures after the failed step closes. A retry rebuilds from the durable log in a new numbered step, success clears the consecutive history, and exhaustion records the structured failure once on `turn/end`. AgentLoop privately owns one cancellation holder whose explicit signal spans prompt policy, assembly, every step, model and tool work, recovery, continuation, and terminal stop; it retires the holder immediately before publishing `turn/end`, while the driver may remain `running` through the durability flush. `cancel()` accepts the typed runtime-only `user | parent` cause, clears pending work, and cooperatively aborts the holder without leaking to the next prompt; durable `turn/end` remains coarse `aborted`, and undispatched model tool calls receive synthetic `tool/call` and `ABORTED_BEFORE_DISPATCH` result pairs. Disposal wins terminal classification, and work that ignores the signal must settle before quiescence. The [explicit-cancellation decision](../../../.agents/notes/implemented/architecture/2026-07-16-explicit-turn-cancellation.md) owns the lifecycle and race contract. Terminal continuation stops remain authoritative through turn close and durability flush.
Plugin failure ends the current turn, not the loop. Only final adapter dispatch/iteration failures and terminal in-band error or aborted finishes enter `agent/request-error`; middleware, result processing, tools, and `agent/post-step` remain ordinary turn failures. Recovery receives the exact live error, immutable provider facts, and immutable prior failures after the failed step closes. A retry rebuilds from the durable log in a new numbered step, success clears the consecutive history, and exhaustion records the structured failure once on `turn/end`. AgentLoop privately owns one cancellation holder whose explicit signal spans prompt policy, assembly, every step, model and tool work, recovery, continuation, and terminal stop; it retires the holder immediately before publishing `turn/end`, while the driver may remain `running` through the durability flush. An effective `cancel()` emits the typed runtime-only `user | parent` cause before clearing pending work and cooperatively aborting the holder; notification failures cannot veto cancellation, work queued by a notification observer is cleared, work queued by a later abort observer belongs to the next turn, and idle cancellation emits nothing. Durable `turn/end` remains coarse `aborted`; undispatched model tool calls receive synthetic `tool/call` and `ABORTED_BEFORE_DISPATCH` result pairs. Disposal wins terminal classification, and work that ignores the signal must settle before quiescence. The [explicit-cancellation decision](../../../.agents/notes/implemented/architecture/2026-07-16-explicit-turn-cancellation.md) owns the lifecycle and race contract. Terminal continuation stops remain authoritative through turn close and durability flush.
Within a step, exclusive calls form barriers; parallel-safe calls use a bounded rolling pool and are reclassified before start. Only dispatch/body overlaps. Policy, durable results, and result context remain model-ordered. Abort stops new calls, drains started results, then drains accepted batch context before the turn closes through the normal abort path.

View File

@@ -325,16 +325,20 @@ export class ReactLoopAgent implements Agent {
}
cancel(cause?: AgentCancelCause): void {
const reason = cause ?? { kind: 'user' }
const resolvedCause = cause ?? { kind: 'user' }
const cancellation = this.turnCancellation
const preRun = cancellation === undefined && (this.#inbox.hasQueued || this.#inbox.hasSteering)
if (preRun) {
this.preRunCancelled = true
if (cancellation !== undefined || preRun) {
if (preRun) this.preRunCancelled = true
// Coordination consumers must update their own state before this call
// clears the inbox or aborts the turn. Notification failures are
// contained by the fused dispatcher and cannot veto cancellation.
agentEvents(this.loopCtx, this).emit('agent/cancel-requested', resolvedCause)
}
// Clear work already present before abort observers run. A replacement
// synchronously enqueued by an observer belongs to the next turn.
this.#inbox.clear()
cancellation?.request(reason)
cancellation?.request(resolvedCause)
}
/**

View File

@@ -6,7 +6,7 @@
* @module dsh-agent-loop/tests/cancel
*/
import { describe, expect, it } from 'vitest'
import { describe, expect, it, vi } from 'vitest'
import { Context } from 'cordis'
import LlmService, { type Message } from '@deepseek-ai/dsh-llm'
import SessionStore, { SessionId, TurnEndReason } from '@deepseek-ai/dsh-session'
@@ -54,6 +54,33 @@ function userTexts(agent: Agent): string[] {
}
describe('Agent.cancel()', () => {
it('notifies every observer before clearing work and contains listener failures', async () => {
const adapter = new MockAdapter([textResponse('must remain unused')])
const ctx = await harness(adapter)
const agent = ctx.agentLoop.create(SessionId('cancel-event'), { provider: 'mock', model: 'mock' })
const warned = vi.spyOn(ctx.logger, 'warn').mockImplementation(() => {})
const seen: string[] = []
ctx.on('agent/cancel-requested', (subject, cause) => {
if (subject !== agent) return
seen.push(`first:${cause.kind}`)
subject.send([{ type: 'text', text: 'queued by cancel observer' }])
throw new Error('observer failed')
})
ctx.on('agent/cancel-requested', (subject, cause) => {
if (subject === agent) seen.push(`second:${cause.kind}`)
})
send(agent, 'drop me')
agent.cancel()
await new Promise(resolve => setTimeout(resolve, 30))
agent.cancel({ kind: 'parent' })
expect(seen).toEqual(['first:user', 'second:user'])
expect(userTexts(agent)).toEqual([])
expect(adapter.requests).toHaveLength(0)
expect(warned).toHaveBeenCalledWith(expect.stringContaining('agent/cancel-requested'))
})
it('cancel() on an idle agent with nothing queued is a no-op; the next prompt runs (F2 leak guard)', async () => {
const adapter = new MockAdapter([textResponse('reply')])
const ctx = await harness(adapter)
@@ -519,7 +546,7 @@ describe('Agent.cancel()', () => {
dispose()
// No step streamed (the model never ran), and the turn ended aborted with
// the CALLER's reason — the marker carries `cancel(reason)` through even
// the caller's cause — the marker carries `cancel(cause)` through even
// though no AbortController observed it in this window.
expect(streamed).toBe(false)
expect(reasons).toEqual([{ kind: 'aborted' }])
@@ -786,7 +813,11 @@ describe('Agent.cancel()', () => {
const flushStarted = Promise.withResolvers<undefined>()
const releaseFlush = Promise.withResolvers<undefined>()
let abortedDuringTurnEnd: boolean | undefined
let cancelNotifications = 0
ctx.on('agent/cancel-requested', (subject) => {
if (subject === agent) cancelNotifications += 1
})
ctx.on('session/event', (session, event) => {
if (session !== agent.session || event.type !== 'turn/end') return
const signal = adapter.requests[0]?.signal
@@ -809,6 +840,7 @@ describe('Agent.cancel()', () => {
expect(abortedDuringTurnEnd).toBe(false)
expect(signal.aborted).toBe(false)
expect(cancelNotifications).toBe(0)
expect(agent.session.events.findLast(event => event.type === 'turn/end')).toMatchObject({
data: { reason: { kind: 'completed' } },
})

View File

@@ -529,7 +529,7 @@ describe('steering from late extension points is never stranded', () => {
expect(JSON.stringify(adapter.requests[1]!.messages)).toContain('one more thing')
})
it('steer() from a step/end session-event listener forces a SAME-TURN next step (/goal pattern)', async () => {
it('steer() from a step/end session-event listener forces a SAME-TURN next step', async () => {
// Assert the same-turn shape; content alone cannot distinguish re-enqueue.
const adapter = new MockAdapter([
textResponse('no tools, would stop'),

View File

@@ -44,7 +44,7 @@ Agent *creation* is provided by the plugin implementing `AgentFactory` (`dsh-age
The lifecycle edges have two important local caveats. `agent/created` runs after scoped setup and after both session and agent registry entries exist. Setup is trusted composition-only code; the immediately following non-vetoing `agent/session-start` notification is the first supported startup injection point. `agent/disposed` always means the exact agent has left the registry. AgentLoop emits it after its driver is quiescent, while ordered teardown may still be detaching the session and unwinding the scope; custom agents registered directly own any stronger driver-ordering contract themselves.
Most interception points are cooperative waterfalls returning seam-specific decisions. Turn-scoped asynchronous seams receive one explicit `AbortSignal`, with `signal` immediately before a waterfall's final `next`; listeners may cooperate but must not retain it as authority over another turn. The signal remains authoritative through terminal policy and is retired immediately before `turn/end` publication, so terminal observers and the following durability flush cannot cancel completed turn work. `agent/pre-step` and `agent/post-step` are serial checkpoints around a step's durable work, while `agent/request-error` is the failed-model-request recovery waterfall: it receives the exact error, normalized failure facts, immutable prior-retried facts, and signal after the failed step closes; a retry opens a new numbered step. `agent/turn-stop` is the terminal serial fold: it runs after ordinary continuation and steering folding, and a returned stop remains in force through turn close and flush so later steering cannot create an extra step or turn. Ordinary queued prompts remain intact. The [explicit-cancellation decision](../../../.agents/notes/implemented/architecture/2026-07-16-explicit-turn-cancellation.md) owns signal lifetime; the [agent-scope runtime-design Agent Note](../../../.agents/notes/implemented/architecture/2026-07-12-agent-scope-runtime-design.md#three-execution-boundaries-are-deliberately-one-way) owns scoped dispatch and terminal settlement.
Most interception points are cooperative waterfalls returning seam-specific decisions. Turn-scoped asynchronous seams receive one explicit `AbortSignal`, with `signal` immediately before a waterfall's final `next`; listeners may cooperate but must not retain it as authority over another turn. The signal remains authoritative through terminal policy and is retired immediately before `turn/end` publication, so terminal observers and the following durability flush cannot cancel completed turn work. `agent/pre-step` and `agent/post-step` are serial checkpoints around a step's durable work, while `agent/request-error` is the failed-model-request recovery waterfall: it receives the exact error, normalized failure facts, immutable prior-retried facts, and signal after the failed step closes; a retry opens a new numbered step. `agent/turn-stop` is the terminal serial fold: it runs after ordinary continuation and steering folding, and a returned stop remains in force through turn close and flush so later steering cannot create an extra step or turn. Ordinary queued prompts remain intact. Effective broad cancellation first emits the observe-only `agent/cancel-requested` with its resolved typed cause, then clears queues and aborts; notification failures are contained and cannot veto the stop. The [explicit-cancellation decision](../../../.agents/notes/implemented/architecture/2026-07-16-explicit-turn-cancellation.md) owns signal lifetime; the [agent-scope runtime-design Agent Note](../../../.agents/notes/implemented/architecture/2026-07-12-agent-scope-runtime-design.md#three-execution-boundaries-are-deliberately-one-way) owns scoped dispatch and terminal settlement.
`PromptDecision.additionalContexts` is an array so every injected context keeps its own source and metadata. A `ContinuationDecision` reason is narrower: it becomes a `steering/message`, not a `context/message`, and therefore carries only content and source.
@@ -54,10 +54,10 @@ Turn and step boundaries and the model token stream are durable `session/event`
The handle every plugin programs against:
- `agent.send(content, options?)` — queue one independent FIFO item. If claimed, that item becomes the sole ordinary message in its turn; a claimed FIFO successor waits for that turn's checkpoint to settle. Broad cancellation, disposal, or a pre-start failure may instead drop it without a turn. Content and resolved source become one detached, deeply frozen lossless-JSON record before `agent/queued` and enqueue; invalid data throws synchronously, and caller or notification-listener in-place mutation cannot change the log or model input (`agent/prompt-submit` still rewrites by returning replacement content). The [one-send-one-turn Agent Note](../../../.agents/notes/implemented/simplification/2026-07-17-one-send-one-turn.md) owns the rationale.
- `agent.send(content, options?)` — queue one independent FIFO item. If claimed, that item becomes the sole ordinary message in its turn; a claimed FIFO successor waits for that turn's checkpoint to settle. Broad cancellation, disposal, or a pre-start failure may instead drop it without a turn. Omitting `options.source` attests direct human input as `{ kind: 'user' }` and may authorize policy consumers, so plugins, schedulers, and other non-human producers provide their own source. Content and resolved source become one detached, deeply frozen lossless-JSON record before `agent/queued` and enqueue; invalid data throws synchronously, and caller or notification-listener in-place mutation cannot change the log or model input (`agent/prompt-submit` still rewrites by returning replacement content). The [one-send-one-turn Agent Note](../../../.agents/notes/implemented/simplification/2026-07-17-one-send-one-turn.md) owns the rationale.
- `agent.steer(content, options?)` — submit steering while the agent is `running`. An open turn records it at the next steering checkpoint before a request or continuation decision; policy can still stop before another step. After turn close and its checkpoint, remaining steering becomes later queued input unless terminal turn policy, cancellation, or disposal discards it. The method uses the same synchronous snapshot-and-validation boundary as `send` and delegates to `send` when idle
- `agent.inject(content, options?)` — accept detached in-session context without running the model; the next request sees its `context/message` with `content` rendered verbatim as a user-role message. `options.meta` persists opaque JSON state without rendering it. While a turn is open it joins that turn, deferring FIFO while the current tool batch executes and draining before turn close if execution is interrupted; while idle it is wrapped in a one-shot `injection` turn and durability checkpoint ([the turn-enclosure invariant](../../../.agents/notes/implemented/architecture/2026-06-15-turn-enclosure-invariant.md)).
- `agent.cancel(cause?)` — cancel ALL pending work: an omitted cause means `{ kind: 'user' }`; TypeScript restricts callers to the `user | parent` union, and an active holder copies its discriminant into a detached frozen signal reason before aborting. The same-process typed seam adds no runtime validation or compatibility fallback for untyped callers. Repeated active-turn cancellation is first-wins, and idle cancellation is a safe no-op that does not arm the next turn. ACP maps to `user`, while in-process parent propagation maps to `parent`. The cause is runtime-only; durable `turn/end` stays coarse `aborted`.
- `agent.cancel(cause?)` — cancel ALL pending work: an omitted cause means `{ kind: 'user' }`; TypeScript restricts callers to the `user | parent` union, and an active holder copies its discriminant into a detached frozen signal reason before aborting. An effective call emits `agent/cancel-requested` with the cause before clearing queued and steering work; observers may synchronize state but cannot veto cancellation. The same-process typed seam adds no runtime validation or compatibility fallback for untyped callers. Repeated active-turn cancellation is first-wins for the signal, and idle cancellation is a safe no-op with no notification. ACP maps to `user`, while in-process parent propagation maps to `parent`. The cause is runtime-only; durable `turn/end` stays coarse `aborted`.
- `agent.whenIdle()` — resolve once the agent reaches quiescence after settling out of `running` (idle → immediately; disposed → awaits the loop exit). A non-owner's quiescence-observation hook: it observes the work settling WITHOUT tearing the agent down. Teardown is separate — a lifecycle owner stops and unregisters via `AgentHandle.dispose()`, which awaits the loop exit directly.
- `agent.session`, `agent.status`, `agent.options`, `agent.id`

View File

@@ -25,7 +25,10 @@ export interface AgentOptions {
model?: string
}
/** Message options; an omitted source resolves to `{ kind: 'user' }`, so plugins must label their own content. */
/**
* Message options. An omitted source attests direct human input as `{ kind: 'user' }`
* and may authorize policy consumers, so non-human producers must label their content.
*/
export interface SendOptions {
source?: MessageSource
}
@@ -130,10 +133,11 @@ export interface Agent {
/**
* Clear all queued and steering work, including items waiting to start, and
* abort the active turn. The first cause wins for that turn, and `whenIdle()`
* resolves after cancellation reaches quiescence. Omission means
* `{ kind: 'user' }`. Idle cancellation is a no-op and does not arm a later
* cancel. The active turn snapshots and freezes the typed cause.
* abort the active turn. An effective call first emits
* `agent/cancel-requested` with the resolved typed cause. The first cause wins
* for the active turn, and `whenIdle()` resolves after cancellation reaches
* quiescence. Omission means `{ kind: 'user' }`. Idle cancellation is a no-op
* and does not arm later work. The active turn snapshots and freezes the cause.
* @param cause - the stable caller intent carried by the current turn signal.
*/
cancel(cause?: AgentCancelCause): void
@@ -185,6 +189,16 @@ declare module 'cordis' {
* @mode emit
*/
'agent/queued'(this: Scoped<Agent>, agent: Agent, content: ContentBlock[], info: { source: MessageSource; steering: boolean }): void
/**
* Effective broad cancellation was requested, before queued/steering work
* is cleared or the active turn is aborted. This observe-only notification
* cannot veto cancellation; listener failures are contained.
* @param agent - the agent whose current work is being cancelled.
* @param cause - resolved typed cancellation cause, including the default.
* Scope-filtered dispatch (`@deepseek-ai/dsh-scope`): agent-scoped listeners receive only that agent.
* @mode emit
*/
'agent/cancel-requested'(this: Scoped<Agent>, agent: Agent, cause: AgentCancelCause): void
// ---- session lifecycle (emit) ----
/**

View File

@@ -188,6 +188,7 @@ describe('agentEvents()', () => {
describe('explicit cancellation helpers', () => {
it('exposes the closed typed cancellation cause at the Agent seam', () => {
expectTypeOf<Parameters<Agent['cancel']>[0]>().toEqualTypeOf<AgentCancelCause | undefined>()
expectTypeOf<Parameters<Events['agent/cancel-requested']>[1]>().toEqualTypeOf<AgentCancelCause>()
})
it('reads only supported reasons from an explicit signal', () => {

View File

@@ -23,7 +23,7 @@ describe('gen-tool-catalog collectToolCatalog', () => {
it('boots every shipped tool package and harvests its model-facing schemas', async () => {
const catalog = await collectToolCatalog()
const names = catalog.flatMap(entry => entry.schemas.map(s => s.name)).sort()
expect(names).toEqual(['ask_user_question', 'bash', 'cordis_inspect', 'cordis_mount', 'cordis_unmount', 'edit', 'glob', 'grep', 'read', 'run_code', 'skill', 'subagent', 'task_kill', 'task_list', 'task_output', 'todo_write', 'web_fetch', 'web_search', 'workflow', 'write'])
expect(names).toEqual(['ask_user_question', 'bash', 'cordis_inspect', 'cordis_mount', 'cordis_unmount', 'create_goal', 'edit', 'get_goal', 'glob', 'grep', 'ralph', 'read', 'run_code', 'skill', 'subagent', 'task_kill', 'task_list', 'task_output', 'todo_write', 'update_goal', 'web_fetch', 'web_search', 'workflow', 'write'])
// Every tool carries a JSON-Schema `parameters` object (what the model sees).
for (const entry of catalog) {
for (const schema of entry.schemas) {