skill-retro

To analyze Claude Code skill performance after a session, run the retro workflow to preprocess the transcript, compare observed behavior against SKILL.md guidance, and generate a structured performance report with remediation findings.

Skill invocation tracking is the process of evaluating trigger accuracy, execution quality, and gap coverage for invoked skills. You need it when you want to identify discrepancies between intended SKILL.md guidance and observed session behavior.

You evaluate trigger accuracy and execution quality by running the retro workflow, which reads SKILL.md files for each invoked skill, compares them against observed session behavior, and outputs structured findings for remediation.

Yes, you can use the post-session review to improve SKILL.md files. The retro workflow surfaces concrete recommendations and action items by comparing intended versus observed behavior to guide skill-creator-driven refinements.

No dependencies are required to run a post-session skill review. The Skill operates independently using its internal scripts and references to preprocess the session transcript and analyze invoked skills.

The best way to find gaps in skill execution coverage is to use an analysis tool that compares intended SKILL.md behavior against actual session performance, generating a comprehensive report with structured findings for remediation.