writing-plans

Implementation plans decompose a feature spec into bite-sized, testable tasks using a DRY approach. This decomposition promotes test-driven development and frequent commits by mapping exact file changes across subsystems.

File-structure mapping identifies exactly which files to create and modify before coding, including where tests and documentation updates belong. This prevents rework by ensuring all file changes are planned before implementation begins.

The review loop takes a saved plan document and validates it against required structure, task granularity rules, and execution handoff instructions. After passing review, the plan is ready for subagent-driven or inline execution.

No TDD experience is required, but implementation plans are TDD-oriented, guiding you to structure work into independently testable tasks. This format helps engineers with little codebase context follow test-first execution steps.

Yes, implementation plans handle multi-subsystem features by ensuring each subsystem gets a separate, independently testable plan. This decomposition maps exact file changes and documentation updates for each subsystem.

Formal implementation plans are unnecessary for simple, single-step changes requiring no multi-step execution or unfamiliar codebase navigation. They are designed for ambiguous specs requiring task decomposition and review-loop handoff.