Humanize Refine Plan ingests an annotated implementation plan (using CMT:/ENDCMT markers) and outputs a cleaned, comment-free plan plus a QA ledger that records every comment, classification, disposition, and any applied changes. It preserves the gen-plan schema and writes outputs atomically, with optional translated variants. Concrete use case: convert an in-repo design draft with reviewer comments into a ready-to-execute plan and a QA audit trail.
Use this skill when you have a plan draft containing inline reviewer comments (CMT blocks) and you need a production-ready plan and a QA ledger documenting decisions. Choose discussion mode for ambiguous comments (prompt the user) or direct mode to auto-resolve with minimal safe assumptions.
Compatible with agent workflows that support scripted flows and local filesystem operations (Claude/Claude-Code/CLI-style agents and automation runners).
This skill has not been reviewed by our automated audit pipeline yet.