meow:task-queue

Prevent file write conflicts during parallel agent execution by enforcing strict file ownership using glob pattern declarations, ensuring only the owning agent can write to specific files. A centralized task queue tracks ownership to avoid cross-task interference.

A centralized task queue for agent orchestration is a JSON-based session-state store that distributes tasks and tracks claims, ownership, and status transitions. It coordinates multiple agents working simultaneously on a shared project.

Agents claim tasks from a shared queue without conflicts by following a strict protocol: checking pending status, verifying no conflicting file ownership, declaring ownership glob patterns, and setting status to in_progress via orchestrator-driven claim serialization.

Yes, you can track task completion history for parallel execution phases. The task queue remains preserved in a JSON session-state file even after tasks are completed, providing a full audit trail and traceability for historical analysis.

Yes, the task queue handles dependencies between parallel tasks. Tasks stored in the session-state JSON file include dependency tracking, allowing the orchestrator to monitor completion and trigger subsequent phases only when dependencies are resolved.

You need strict file ownership for multi-agent collaboration when multiple agents work simultaneously on a shared project containing source files and assets, requiring coordinated task claims to prevent cross-task file interference.