phase-planner

To split approved tasks into executable phases, you decompose them into a deterministic sequence and store the plan in a canonical phases.json file. This enforces requirements like allowed_write_paths, test_policy, and deliverables for reliable task execution.

A canonical phases.json file serves as the standard execution plan for task management. It stores decomposed phase sequences, synchronizes with spec and intake data, and defines required fields like completion_signal and starting_points for reliable planning.

After task approval, run the plan command to generate phases.json and populate it with phase requirements. Then set up bootstrap metadata and transition to the phase-executor to begin the workflow.

Yes, you can synchronize intake and spec data within your phase planning workflow. Storing execution plans in phases.json maintains consistency and ensures intake and spec remain synchronized during task decomposition.

If you need to change an existing phase plan, you must follow a re-locking and re-planning procedure. This ensures the phases.json canonical plan stays synchronized and deterministic before you resume execution.