writing-plans

Provide the spec or requirements, and the Skill produces a plan with a goal, architecture summary, tech stack, and tasks broken into 2-5 minute TDD steps. Each task lists exact files to create or modify, complete code, test commands with expected output, and commit steps.

Each step should be one action taking 2-5 minutes: write the failing test, run it to confirm failure, implement minimal code, verify it passes, then commit. Tasks must include exact file paths and complete code with no placeholders like TODO or TBD.

Split when the spec covers multiple independent subsystems. Each plan should produce working, testable software on its own, so one plan per subsystem keeps scope manageable and execution verifiable.

After saving the plan, you choose between subagent-driven execution, which dispatches a fresh subagent per task with review between tasks, or inline execution, which runs tasks in the current session in batches with checkpoints.

Common failures include placeholder steps, vague instructions like add error handling, references to undefined functions, and inconsistent naming across tasks. The Skill's self-review checklist catches spec gaps, placeholders, and type inconsistencies before handoff.