writing-plans

Convert specs into execution plans by decomposing them into atomic tasks with exact file paths, complete code samples, runnable tests, and verification gates. The plan outputs step-by-step instructions ready for LLM agent handoff, including commit messages and verification commands.

Execution plans must specify exact file paths, complete runnable code, verification commands with expected outputs, and atomic task breakdowns. This ensures agents have unambiguous instructions for implementation, testing, and progress verification across all planning phases.

Yes. Execution plans with precise paths, full code samples, explicit verification commands, and clear task decomposition are designed for agent handoff. The structured format eliminates ambiguity and enables agents to execute and validate work independently.

A complete plan includes exact file paths for every change, minimal but runnable implementation code, test code that validates behavior, and verification gates with specific commands and expected outputs. This completeness prevents execution errors and enables independent agent verification.

Verification gates define explicit commands to run and their expected outputs at each plan phase. They act as checkpoints ensuring tasks complete correctly before proceeding, enabling agents to validate progress and catch issues early in execution.

Execution plans apply only to approved, complete specs. Partial or incomplete specifications lack the detail needed to generate exact paths, full code samples, and verification steps, so they must be finalized before plan generation.