Implements multi-step development plans through systematic batch execution, critical review checkpoints, and automated verification.
The executing-plans skill provides a disciplined framework for Claude to carry out complex development tasks from a written implementation plan. It moves away from 'one-shot' coding by enforcing a batch-based workflow where the AI executes three tasks at a time, runs verifications, and pauses for human feedback. This ensures that architectural integrity is maintained and that any deviations or blockers are addressed immediately rather than compounding errors over a long session.
主な機能
01Strict verification requirements for every task before moving forward
02Systematic batch execution with mandatory human review checkpoints
030 GitHub stars
04Seamless integration with git worktrees for isolated development environments
05Pre-implementation critical review to identify plan gaps or blockers
06Granular task tracking with in-progress and completion status markers
ユースケース
01Executing large-scale refactors documented in a technical specification
02Building complex features that require human verification at specific milestones
03Methodically addressing a backlog of tasks while maintaining code quality and safety