skill-retro

Analyze skill execution logs to catalog corrections and update SKILL.md files.

9|2|Updated Mar 4, 2026
One-click install
npx skills add https://github.com/typemd/typemd --skill skill-retro-typemd
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: skill-retro
Source: https://github.com/typemd/typemd/tree/main/.claude/skills/skill-retro
Command: npx skills add https://github.com/typemd/typemd --skill skill-retro-typemd

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Skill‑retro analyzes a conversation’s skill execution history to spot interruptions, self‑corrections, and unnecessary pauses, then automatically updates the offending SKILL.md and related metadata so future runs are smoother.

Core Features & Use Cases

  • Identify which skill(s) were invoked and rank them by correction frequency.
  • Catalog each correction event with type, location, expected vs actual behavior, and root cause.
  • Propose and apply targeted edits to SKILL.md, including script extraction suggestions.
  • Verify the updated skill for consistency and validity after changes.
  • Ideal for developers iterating on Claude skills who want continuous improvement without manual review.

Quick Start

Invoke /skill-retro to analyze the latest session and automatically refine the relevant skill's definition.

Frequently Asked Questions about skill-retro

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I automatically analyze skill execution logs to find interruptions and self-corrections?

You can use skill-retro to analyze recent skill execution logs, identify interruptions, and catalog self-correction events by parsing conversation history for tool calls. It targets invoked skills, ranks them by correction frequency, and generates concrete updates to their SKILL.md files.

What is the best way to update SKILL.md files automatically after a Claude skill fails or pauses unnecessarily?

The best way to update SKILL.md files automatically is to run a retroanalysis on the session. This process catalogs each correction event with its root cause and applies targeted edits, including script extraction suggestions, directly to the skill metadata.

How do I catalog correction events and root causes from conversation history?

To catalog correction events from conversation history, you need to parse the tool calls within the session logs. The analysis identifies the expected versus actual behavior, locates the correction event, and determines the root cause for each invoked skill.

Do I need read and write access to SKILL.md files to apply corrective fixes?

Yes, you need read and write access to SKILL.md files. This access is required to parse conversation history for tool calls, generate concrete updates to the skill definitions, and verify the updated skill for consistency and validity after changes.

Can I verify the consistency of a skill definition after applying automated metadata corrections?

Yes, you can verify the consistency of a skill definition after applying automated metadata corrections. The analysis process validates the updated skill files after proposing and applying targeted edits to ensure future runs are smoother.

When should I run a retroanalysis on my Claude skills to improve execution?

You should run a retroanalysis when developers iterate on Claude skills and want continuous improvement without manual review. It is ideal for spotting unnecessary pauses and self-corrections, then automatically refining the relevant skill's definition.