decompose

To break a project brief into parallel tasks, the Skill parses the brief, scans the codebase for module boundaries and data dependencies, and generates parallelizable Track task files along with a saved decomposition plan. This ensures AI agents can execute repository changes without overlapping or sequentially blocked work.

Task decomposition handles multi-module repositories by exploring the codebase to map safe file-scope splits and dependency chains. It identifies boundaries that prevent overlapping work, ensuring generated tasks remain parallel-safe across modules even when file scopes intersect.

Yes, you can regenerate a task breakdown after a brief changes. The Skill performs re-decomposition by preserving active or completed work, canceling stale todo items, running Track validation, and offline regenerating Track views to keep the task graph consistent with the updated brief.

The best way to clean up stale todo items during re-decomposition is through automated task lifecycle management. The Skill preserves active or completed work, cancels outdated stale tasks automatically, creates new task files, and saves the updated decomposition plan.

Parallel task generation works with existing codebases that need fresh decomposition or re-decomposition. The Skill requires codebase exploration to find module boundaries, data dependencies, and safe file-scope splits, making it suitable for evolving multi-module repositories rather than just new projects.

You need automated project decomposition when a product brief must be broken into agent-ready tasks for a multi-module codebase, especially when tasks must be parallel-safe and overlap-free. It is required when briefs evolve and stale tasks need cleanup to maintain a consistent task graph.