What problem does it solve? Before executing an implementation plan, it is hard to spot your own blind spots. This Skill gets an independent second opinion by sending your plan to a sandboxed Claude Code subprocess that grades it and lists concrete concerns. ## Core Features & Use Cases - Independent Plan Review: Spawns a claude -p subprocess with tools disabled so the review is analysis-only and cannot execute code or read files. - Structured Feedback: Returns a grade from 0 to 5, an overall assessment, severity-tagged concerns ([Critical], [Important], [Minor]), and a summary of what works well. - Resumable Sessions: Captures the subprocess session ID so you can continue the review conversation with follow-up questions via --resume. - Use Case: You just finished drafting an implementation plan in ~/.claude/plans/. Run the Skill to have a senior-engineer persona check it for missing edge cases, security implications, and dependency ordering problems before you start coding. ## Quick Start Ask the AI to review your latest implementation plan, or provide a specific path such as /review-plan ~/plans/my-plan.md to review a particular file.