What problem does it solve? After a working session, it is hard to know whether the right skills fired, which ones were missed, and where the fix belongs. This Skill reconstructs the trigger record from the session log and skill files on disk, classifies every miss by the channel that failed, and routes each vetted finding to the surface that owns the repair. ## Core Features & Use Cases - Trigger record reconstruction: Lists which skills actually loaded during the session and verifies each cited skill's frontmatter on disk before judging its trigger text. - Five finding classes: Categorizes outcomes as fired-and-earned, should-have-fired, fired-wrongly, fired-and-ignored, or uncovered gap, each mapped to the trigger channel that failed. - Owner-based routing: Sends defects in sd-* skills to the skills/ directory, files work items against other repositories, and forwards uncovered gaps to sd-propose-skills. - Use Case: After a long development session, ask for a skill retro to discover that a planning skill never fired because its description wording missed the opening request, then get a vetted, routed fix proposal. ## Quick Start Ask the AI to run a skill retro on the session that just finished, reviewing which skills fired, which were missed, and where each fix should be filed.