writing-plans

To break down complex software features, you create a detailed implementation plan that decomposes them into 2-5 minute tasks with exact file paths, tests, and minimal code. This ensures clear task boundaries, acceptance criteria, and a repeatable review process.

The best way to plan implementation for multi-agent development is to scope plans for multi-step features and delegation to subagents, documenting architecture choices and verification steps to prove each task can be completed independently and verified by tests and commits.

You enforce TDD-style iteration by generating a structured plan that includes explicit file paths, tests, and commit steps for each decomposed task. This enforces test-driven development cycles and ensures every unit of work is verified independently.

Yes, planning subagent development requires exact file paths, tests, and acceptance criteria. Scoping plans with these elements ensures clear task boundaries, allowing subagents to complete tasks independently and verify them through tests and commits.

You need to document architecture choices and trade-offs during task decomposition when planning complex features and design-to-implementation handoffs. This documentation provides a repeatable template that proves each task can be completed and verified independently.