Commit Graph

325 Commits

Author SHA1 Message Date
tad-hq 5d5755f29d fix(circuit-breaker): wire target-aware detection into background manager 2026-03-17 13:40:46 -06:00
tad-hq 1fdce01fd2 fix(circuit-breaker): target-aware loop detection via tool signatures 2026-03-17 13:36:09 -06:00
tad-hq c8213c970e fix(circuit-breaker): add enabled config flag as escape hatch 2026-03-17 13:29:06 -06:00
YeonGyu-Kim b788586caf relax task timeouts: stale timeout 3min→20min, session wait 30s→1min 2026-03-17 16:47:13 +09:00
YeonGyu-Kim 0b4d092cf6 Merge pull request #2639 from code-yeongyu/feature/2635-smart-circuit-breaker
feat(background-agent): add smart circuit breaker for repeated tool calls
2026-03-17 16:43:08 +09:00
YeonGyu-Kim 53285617d3 Merge pull request #2636 from code-yeongyu/fix/pre-publish-blockers
fix: resolve 12 pre-publish blockers (security, correctness, migration)
2026-03-17 16:36:04 +09:00
YeonGyu-Kim ae3befbfbe fix(background-agent): apply smart circuit breaker to manager events
Ultraworked with [Sisyphus](https://github.com/code-yeongyu/oh-my-openagent)

Co-authored-by: Sisyphus <clio-agent@sisyphuslabs.ai>
2026-03-17 16:31:55 +09:00
YeonGyu-Kim dc1a05ac3e feat(background-agent): add loop detector helpers
Ultraworked with [Sisyphus](https://github.com/code-yeongyu/oh-my-openagent)

Co-authored-by: Sisyphus <clio-agent@sisyphuslabs.ai>
2026-03-17 16:31:55 +09:00
YeonGyu-Kim 2c7ded2433 fix(background-agent): defer task cleanup while siblings running 2026-03-17 15:17:34 +09:00
MoerAI 3055454ecc fix(background-agent): add circuit breaker to prevent subagent infinite loops
Adds a configurable maxToolCalls limit (default: 200) that automatically
cancels background tasks when they exceed the threshold. This prevents
runaway subagent loops from burning unlimited tokens, as reported in #2571
where a Gemini subagent ran 809 consecutive tool calls over 3.5 hours
costing ~$350.

The circuit breaker triggers in the existing tool call tracking path
(message.part.updated/delta events) and cancels the task with a clear
error message explaining what happened. The limit is configurable via
background_task.maxToolCalls in oh-my-opencode.jsonc.

Fixes #2571
2026-03-16 11:07:33 +09:00
YeonGyu-Kim 786c7a84d0 fix(background-agent): prevent queue item loss on concurrent cancel and guard against cancelled task resurrection 2026-03-13 13:12:59 +09:00
YeonGyu-Kim fd71c89b95 fix(background-agent): release descendant quota on pre-start task cancellation and creation failure 2026-03-13 12:37:33 +09:00
YeonGyu-Kim 457f303adf fix(background-agent): clean global subagentSessions and SessionCategoryRegistry on dispose 2026-03-13 10:56:44 +09:00
YeonGyu-Kim 4157c2224f fix(background-agent): clear pending parent on silent cancel
Ultraworked with [Sisyphus](https://github.com/code-yeongyu/oh-my-openagent)

Co-authored-by: Sisyphus <clio-agent@sisyphuslabs.ai>
2026-03-12 11:04:35 +09:00
YeonGyu-Kim 57b4985424 fix(background-agent): delay session error task cleanup
Ultraworked with [Sisyphus](https://github.com/code-yeongyu/oh-my-openagent)

Co-authored-by: Sisyphus <clio-agent@sisyphuslabs.ai>
2026-03-12 02:24:42 +09:00
YeonGyu-Kim cf276322a3 fix(background-agent): handle async shutdown in process-cleanup signal handlers 2026-03-12 01:37:03 +09:00
YeonGyu-Kim 2c3c447dc4 fix: address review-work round 3 findings (async shutdown, signal generation, stale test name) 2026-03-12 01:37:03 +09:00
YeonGyu-Kim ff536e992a fix: address review-work round 2 findings
- MCP teardown race: add shutdownGeneration counter to prevent
  in-flight connections from resurrecting after disconnectAll
- MCP multi-key disconnect race: replace disconnectedSessions Set
  with generation-based Map to track per-session disconnect events
- MCP clients: check shutdownGeneration in stdio/http client
  creators before inserting into state.clients
- BackgroundManager: call clearTaskHistoryWhenParentTasksGone after
  timer-based task removal in scheduleTaskRemoval and notifyParentSession
- BackgroundManager: clean completedTaskSummaries when parent has
  no remaining tasks
- Plugin dispose: remove duplicate tmuxSessionManager.cleanup call
  since BackgroundManager.shutdown already handles it via onShutdown
2026-03-12 01:37:03 +09:00
YeonGyu-Kim 03eaa429ce fix: address 5 edge cases from review-work findings
- C3: include command args in auto-slash-command dedup key
- H2: track completed task summaries for ALL COMPLETE message
- H9: increment tmux close retry count on re-mark
- H8: detect stale MCP connections after disconnect+reconnect race
- H8: guard disconnectedSessions growth for non-MCP sessions
- C1: await tmux cleanup in plugin dispose lifecycle
2026-03-12 01:37:03 +09:00
YeonGyu-Kim b8aea50dfa test(background-agent): update completion timer test for per-task cleanup
Test expected timers only after allComplete, but H2 fix intentionally
decoupled per-task cleanup from sibling completion state. Updated
assertion to expect timer after individual task notification.
2026-03-12 01:37:03 +09:00
YeonGyu-Kim b4e13883b1 fix(background-agent): fix 3 memory leaks in task lifecycle management
H3: cancelTask(skipNotification=true) now schedules task removal.
Previously the early return path skipped cleanup, leaking task objects
in this.tasks Map permanently. Extracted scheduleTaskRemoval() helper
called from both skipNotification and normal paths.

H2: Per-task completion cleanup timer decoupled from allComplete check.
Previously cleanup timer only ran when ALL sibling tasks completed. Now
each finished task gets its own removal timer regardless of siblings.

H1+C2: TaskHistory.clearAll() added and wired into shutdown(). Added
clearSession() calls on session error/deletion and prune cycles.
taskHistory was the only data structure missed by shutdown().

Tests: 10 pass (3 cancel + 3 completion + 4 history)
2026-03-12 01:37:03 +09:00
YeonGyu-Kim aaa54858a3 fix(background-agent): extend default no-progress stale timeout to 30 minutes
Give never-updated background tasks a longer default window and keep the default-threshold regression coverage aligned with that behavior.

Ultraworked with [Sisyphus](https://github.com/code-yeongyu/oh-my-openagent)

Co-authored-by: Sisyphus <clio-agent@sisyphuslabs.ai>
2026-03-12 01:14:35 +09:00
YeonGyu-Kim 7dddf99d9a Merge pull request #2470 from code-yeongyu/fix/terminal-task-retention-ttl
fix(background-agent): add TTL for terminal task retention
2026-03-11 21:57:33 +09:00
YeonGyu-Kim 0d9f001c11 fix(background-agent): add TTL for terminal task retention to prevent unbounded growth
Ultraworked with [Sisyphus](https://github.com/code-yeongyu/oh-my-openagent)

Co-authored-by: Sisyphus <clio-agent@sisyphuslabs.ai>
2026-03-11 21:52:48 +09:00
YeonGyu-Kim cc1c23032f fix(background-agent): handle SDK error response in spawn limit lineage lookup
- Check response.error and !response.data after session.get() to fail closed
- Prevents unlimited spawning when SDK returns non-throwing error responses
- Added regression tests for SDK error and missing data scenarios
2026-03-11 21:45:40 +09:00
YeonGyu-Kim 4b5c47172d Merge pull request #2449 from code-yeongyu/fix/issue-2330-v2
fix(background-agent): cap recursive subagent spawning
2026-03-11 21:00:34 +09:00
YeonGyu-Kim 594233183b fix(background-agent): fail closed on spawn lineage lookup errors
Ultraworked with [Sisyphus](https://github.com/code-yeongyu/oh-my-openagent)

Co-authored-by: Sisyphus <clio-agent@sisyphuslabs.ai>
2026-03-11 20:57:09 +09:00
YeonGyu-Kim e3b17da4bd fix(background-agent): preserve terminal tasks until notification cleanup
Route terminal task cleanup through parent notifications so cancelled and errored tasks stay visible until delayed cleanup finishes.

Ultraworked with [Sisyphus](https://github.com/code-yeongyu/oh-my-openagent)

Co-authored-by: Sisyphus <clio-agent@sisyphuslabs.ai>
2026-03-11 20:50:49 +09:00
YeonGyu-Kim 8c5f9b8082 fix(background-agent): skip terminal tasks during stale pruning
Prevent TTL pruning from deleting terminal tasks before delayed notification cleanup runs.

Ultraworked with [Sisyphus](https://github.com/code-yeongyu/oh-my-openagent)

Co-authored-by: Sisyphus <clio-agent@sisyphuslabs.ai>
2026-03-11 20:50:49 +09:00
YeonGyu-Kim a179ebe0b9 Count sync subagent spawns against descendant limits 2026-03-11 20:50:11 +09:00
YeonGyu-Kim 4a39c83eb5 Limit recursive subagent spawning 2026-03-11 20:50:10 +09:00
YeonGyu-Kim 3550305af8 Merge branch 'dev' into fix/issue-2232 2026-03-11 20:20:04 +09:00
YeonGyu-Kim 413e8b73b7 Add session permission support to background agents for denying questions
Implements question-denied session permission rules when creating child
sessions via background task delegation. This prevents subagents from
asking questions by passing explicit permission configuration during
session creation.

🤖 GENERATED WITH ASSISTANCE OF OhMyOpenCode
2026-03-11 19:42:46 +09:00
YeonGyu-Kim 1120885fd0 fix(background-agent): release interrupted task slots during startup cleanup
Ultraworked with [Sisyphus](https://github.com/code-yeongyu/oh-my-opencode)

Co-authored-by: Sisyphus <clio-agent@sisyphuslabs.ai>
2026-03-09 12:36:49 +09:00
YeonGyu-Kim 85aa744c8a fix(background-agent): clear toast tracking when tasks stop
🤖 Generated with assistance of [OhMyOpenCode](https://github.com/code-yeongyu/oh-my-opencode)
2026-03-09 11:39:04 +09:00
YeonGyu-Kim beb89faa0f Merge pull request #2388 from code-yeongyu/fix/background-output-undefined-status-2387
fix: handle undefined sessionStatus in pollRunningTasks (#2387)
2026-03-08 23:54:58 +09:00
YeonGyu-Kim dc370f7fa8 fix: handle undefined sessionStatus in pollRunningTasks (#2387)
When a completed session is no longer returned by session.status(),
allStatuses[sessionID] is undefined. Previously this fell through to
a 'still running' log, leaving the task stuck as running forever.

Match the sync-session-poller pattern: only continue (skip completion
check) when sessionStatus EXISTS and is not idle. When undefined,
fall through to validateSessionHasOutput + checkSessionTodos +
tryCompleteTask, same as idle.
2026-03-08 23:42:11 +09:00
YeonGyu-Kim 2b5dec5333 fix(background-agent): use compaction-aware prompt context in manager
Ultraworked with [Sisyphus](https://github.com/code-yeongyu/oh-my-opencode)

Co-authored-by: Sisyphus <clio-agent@sisyphuslabs.ai>
2026-03-08 02:23:33 +09:00
YeonGyu-Kim c789baf1d9 fix(background-agent): merge prompt context across compaction gaps
Ultraworked with [Sisyphus](https://github.com/code-yeongyu/oh-my-opencode)

Co-authored-by: Sisyphus <clio-agent@sisyphuslabs.ai>
2026-03-08 02:23:27 +09:00
YeonGyu-Kim fade6740ae chore: update GPT-5.2 references to GPT-5.4
Align runtime defaults, tests, docs, and generated artifacts with the newer GPT-5.4 baseline. Keep think-mode and prompt-routing expectations consistent after the model version bump.
2026-03-07 05:46:05 +09:00
YeonGyu-Kim 23dcd99c9a docs(agents): refresh generated AGENTS guides
🤖 GENERATED WITH ASSISTANCE OF [OhMyOpenCode](https://github.com/code-yeongyu/oh-my-opencode)
2026-03-06 17:59:05 +09:00
YeonGyu-Kim 33d39597ae docs(agents): regenerate AGENTS.md hierarchy with updated metrics and model configs
- 1208→1243 TS files (+35), 143k→155k LOC (+12k)
- Update all agent models: Sisyphus, Hephaestus, Oracle, Librarian, Atlas, Metis, Momus
- Add 6 new hook directories (39→45 dirs): beast-mode-system, hashline-edit-diff-enhancer, anthropic-image-context, task-reminder, compaction-todo-preserver, runtime-fallback
- Update category models: visual-engineering/artistry gemini-3-pro→gemini-3.1-pro
- Add 2 config schema files: fallback-models.ts, runtime-fallback.ts
- Timestamp: 2026-03-02 | Commit: 1c2caa09

🤖 Generated with assistance of [OhMyOpenCode](https://github.com/code-yeongyu/oh-my-opencode)
2026-03-02 23:40:38 +09:00
YeonGyu-Kim 050b93bebb fix(background-agent): clean pendingNotifications on session.deleted 2026-02-27 03:00:39 +09:00
YeonGyu-Kim d425f9bb80 fix(models): update Gemini 3 to 3.1 Pro and add Kimi to writing category fallback
Closes #2065

Closes #1968
2026-02-26 21:01:26 +09:00
YeonGyu-Kim 840c612be8 fix(background-agent): queue notifications for idle parent sessions
When a background task completes and the parent session is waiting for
user input, promptAsync() fails with an aborted error. Previously the
notification was silently dropped — lost forever.

Fix: queue the notification text in-memory on the BackgroundManager
when promptAsync fails with an aborted/idle error. On the user's next
message to that session, the queued notifications are injected into the
chat context before the agent sees the message.

- BackgroundManager: add pendingNotifications map + queuePendingNotification()
  and injectPendingNotificationsIntoChatMessage() methods
- background-notification hook: add chat.message handler that calls injection
- chat-message.ts: wire backgroundNotificationHook.chat.message into the
  message processing chain
- Add tests covering queue-on-abort and next-message delivery
2026-02-25 16:26:31 +09:00
YeonGyu-Kim 73453a7191 docs(agents): update hook counts 44→46, add hashline-edit documentation
- Update root AGENTS.md: hook count 44→46, commit fcb90d92, generated 2026-02-24
- Update src/AGENTS.md: core hooks 35→37, session hooks 21→23
- Update src/hooks/AGENTS.md: 46 hooks total, add modelFallback/noSisyphusGpt/noHephaestusNonGpt/runtimeFallback, jsonErrorRecovery moved to tool-guard (tier 2)
- Create src/tools/hashline-edit/AGENTS.md (93 lines): documents three-op model, LINE#ID format, execution pipeline
- Refresh timestamps: 2026-02-21→2026-02-24 on 28 files
- Update plugin/AGENTS.md hook composition counts

🤖 Generated with assistance of [OhMyOpenCode](https://github.com/code-yeongyu/oh-my-opencode)
2026-02-25 00:02:05 +09:00
YeonGyu-Kim 2428a46e6d refactor(features/background-agent): remove dead code submodules
Delete 15 unused files in background-agent module:

- background-task-completer.ts

- format-duration.ts

- message-dir.ts

- parent-session-context-resolver.ts

- parent-session-notifier.ts (and its test file)

- result-handler-context.ts

- result-handler.ts

- session-output-validator.ts

- session-task-cleanup.ts

- session-todo-checker.ts

- spawner/background-session-creator.ts

- spawner/concurrency-key-from-launch-input.ts

- spawner/spawner-context.ts

- spawner/tmux-callback-invoker.ts

Update index.ts barrel and manager.ts/spawner.ts imports

Ultraworked with [Sisyphus](https://github.com/code-yeongyu/oh-my-opencode)

Co-authored-by: Sisyphus <clio-agent@sisyphuslabs.ai>
2026-02-23 02:43:01 +09:00
YeonGyu-Kim a3e4f904a6 refactor(background-agent): wire session-idle-event-handler into manager, add unit tests
The extracted handleSessionIdleBackgroundEvent was never imported by
manager.ts — dead code from incomplete refactoring (d53bcfbc). Replace
the inline session.idle handler (58 LOC) with a call to the extracted
function, remove unused MIN_IDLE_TIME_MS import, and add 13 unit tests
covering all edge cases.
2026-02-22 15:30:40 +09:00
YeonGyu-Kim d1e5bd63c1 fix: address Oracle + Cubic review feedback for background-agent refactoring
- Revert getMessageDir to original join(MESSAGE_STORAGE, sessionID) behavior
- Fix dead subagentSessions.delete by capturing previousSessionID before tryFallbackRetry
- Add .unref() to process cleanup setTimeout to prevent 6s hang on Ctrl-C
- Add missing isUnstableAgent to fallback retry input mapping
- Fix process-cleanup tests to use exit listener instead of SIGINT at index 0
- Swap test filenames in compaction-aware-message-resolver to exercise skip logic correctly
2026-02-22 12:14:26 +09:00
YeonGyu-Kim 8d66d5641a test(background-agent): add unit tests for extracted modules
Add 104 new tests across 4 test files:
- error-classifier.test.ts (80 tests): isRecord, isAbortedSessionError, getErrorText, extractErrorName, extractErrorMessage, getSessionErrorMessage
- fallback-retry-handler.test.ts (19 tests): retry logic, fallback chain, concurrency release, session abort, queue management
- process-cleanup.test.ts (7 tests): signal registration, multi-manager shutdown, cleanup on unregister
- compaction-aware-message-resolver.test.ts (13 tests): compaction agent detection, message resolution with temp dirs (pre-existing, verified)

Total background-agent tests: 161 -> 265 (104 new, 0 regressions)
2026-02-22 11:59:06 +09:00