What problem does it solve? Turning a written plan or spec into working code often goes wrong through skipped reviews, ambiguous requirements, silent long coding stretches, and untested pushes. This Skill enforces a disciplined implementation workflow so plan execution stays reviewed, tested, and communicated at every step. ## Core Features & Use Cases - Pre-implementation review: Reads the full plan, runs a 3-phase code review, cross-references plan vs. codebase, and resolves all ambiguities before any code is written. - Per-chunk approval workflow: Decomposes the plan into small testable chunks, presenting each with a code plan and a local-test vs. CI test split recommendation for explicit approval. - Structured check-ins and CI awareness: Uses a fixed check-in format after every chunk, pushes with plan-referenced commit messages, monitors GitHub Actions status, and hard-stops on failures. - Use Case: You hand the AI a design doc for a new feature and say "implement this" — it reviews the codebase, updates CLAUDE.md/ARCH.md/TODO.md, then implements chunk by chunk with your approval at each stage. ## Quick Start Implement the feature described in docs/plan.md, reviewing the codebase first and checking in with me after each chunk.