Standalone agents (no initiative or 0 linked projects) run in a
workspace/ subdirectory, but signal.json lookups used the parent
directory. This caused all standalone agents to be marked "crashed"
despite successful completion.
Track the actual agent cwd at spawn time via ActiveAgent.agentCwd
and probe for the workspace/ subdirectory during reconciliation and
crash detection paths.
- INTER_AGENT_COMMUNICATION constant → buildInterAgentCommunication(agentId) function
- Manager injects actual agent ID into prompt after DB record creation
- Agent ID hardcoded in cw listen/ask commands — no manifest.json indirection
- cw listen now uses onPendingConversation SSE subscription instead of polling
- CLI trpc-client upgraded with splitLink for subscription support
- All CLI flags (--agent-id, --from, --timeout, --poll-interval) documented in prompt
- conversation:created/answered added to ALL_EVENT_TYPES
Enables parallel agents to communicate through a CLI-based conversation
mechanism coordinated via tRPC. Agents can ask questions to peers and
receive answers, with target resolution by agent ID, task ID, or phase ID.
DB log chunk insertion is now the sole trigger for agent:output events.
Eliminates triple emission (FileTailer, handleStreamEvent, output buffer)
in favor of: FileTailer.onRawContent → DB insert → EventBus emit.
- createLogChunkCallback emits agent:output after successful DB insert
- spawnInternal now wires onRawContent callback (fixes session 1 gap)
- Remove eventBus from FileTailer (no longer touches EventBus)
- Remove eventBus from ProcessManager constructor (dead parameter)
- Remove agent:output emission from handleStreamEvent text_delta
- Remove outputBuffers map and all buffer helpers from manager/handler
- Remove getOutputBuffer from AgentManager interface and implementations
- getAgentOutput tRPC: DB-only, no file fallback
- onAgentOutput subscription: no initial buffer yield, events only
- AgentOutputViewer: accumulates raw JSONL chunks, parses uniformly
PROBLEM:
- Agents completing with questions were incorrectly marked as "crashed"
- Race condition: polling handler AND crash handler both called handleCompletion()
- Caused database corruption and lost pending questions
SOLUTION:
- Add completion mutex in OutputHandler to prevent concurrent processing
- Remove duplicate completion call from crash handler
- Only one handler executes completion logic per agent
TESTING:
- Added mutex-completion.test.ts with 4 test cases
- Verified mutex prevents concurrent access
- Verified lock cleanup on exceptions
- Verified different agents can process concurrently
FIXES: residential-cuckoo and 12+ other agents stuck in crashed state
Replaces file completion detection with a superior approach that reads only
complete JSONL lines and tracks file position. This eliminates race conditions
without any delays or polling.
Key improvements:
- Read up to last complete line, avoiding partial lines during writes
- Track file position per agent for incremental reading
- Process only valid, complete JSON lines
- Clean up position tracking on completion/crash
- No hardcoded delays or polling required
This approach is more robust, responsive, and elegant than timing-based solutions.
The race condition where agents were marked as crashed is now completely resolved.
Fixes race condition where agents were incorrectly marked as crashed when
output files took longer than 500ms to complete writing.
Changes:
- Replace hardcoded 500ms delay with polling-based file completion detection
- Add signal file validation to ensure JSON is complete before processing
- Make status updates atomic to prevent race conditions
- Update cleanup manager to pass outputFilePath for proper timing
This resolves the issue where successful agents like "abundant-wolverine"
were marked as crashed despite producing valid output.
- Add structured_output field to ClaudeCliResult interface
- Read from structured_output when present (--json-schema response)
- Fall back to parsing result for backwards compatibility
- Import all mode-specific JSON schemas (discuss, breakdown)
- Add getJsonSchemaForMode() helper to select schema by mode
- Update spawn() to pass mode to repository and use mode-specific schema
- Refactor handleAgentCompletion() to route to mode-specific handlers
- Add handleExecuteOutput() for execute mode (existing behavior)
- Add handleDiscussOutput() for discuss mode (context_complete status)
- Add handleBreakdownOutput() for breakdown mode (breakdown_complete status)
- Update resume() to use mode-specific JSON schema
- Add AgentMode type: 'execute' | 'discuss' | 'breakdown'
- Add mode column to agents table with 'execute' default
- Update SpawnAgentOptions to accept optional mode
- Update AgentInfo interface to include mode field
- Update ClaudeAgentManager.toAgentInfo to map mode
- Fix MockAgentManager to include mode in spawn
- Fix dispatch manager tests to include mode
- Change resume() signature from (agentId, prompt) to (agentId, answers)
- Accept Record<string, string> mapping question IDs to user answers
- Format answers as structured prompt for Claude CLI
- Update AgentManager interface in types.ts
- Update manager tests for new signature
- Change status from 'question' to 'questions' (plural)
- Add QuestionItem with id field for answer matching
- Update PendingQuestion to PendingQuestions with questions array
- Update AgentWaitingEvent payload to questions array
- Update ClaudeAgentManager and MockAgentManager adapters
- Update TestHarness and all test files
- Pass --json-schema flag to Claude CLI for validated output
- Parse discriminated union (done/question/unrecoverable_error) in handleAgentCompletion
- Add getPendingQuestion method to AgentManager interface
- Add PendingQuestion type for structured question data
- Store pending question in ActiveAgent for later retrieval
- Remove hacky string matching for waiting_for_input detection
- Update MockAgentManager with getPendingQuestion and options support
- Update tests for new CLI arguments and result format
- Test spawn with worktree and agent record creation
- Test duplicate name rejection
- Test AgentSpawned event emission
- Test stop with subprocess kill and status update
- Test list, get, getByName operations
- Test resume with session_id and --resume flag
- Test AgentResumed event emission
- Fix: use agent.id from repository for activeAgents tracking
- Use Claude CLI with --output-format json for agent spawning
- Extract session_id from JSON result for resume capability
- Emit lifecycle events: spawned, stopped, crashed, resumed, waiting
- Handle waiting_for_input status for AskUserQuestion pauses
- Uses WorktreeManager for isolated agent workspaces
