retro

Analyze Claude Code session transcripts to detect friction and route findings.

3|Updated May 12, 2026
One-click install
npx skills add https://github.com/netresearch/retro-skill --skill retro-netresearch
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: retro
Source: https://github.com/netresearch/retro-skill/tree/main/skills/retro
Command: npx skills add https://github.com/netresearch/retro-skill --skill retro-netresearch

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill requires python3, jq, gh, glab, git, and includes scripts (resource) and references (resource) components.

What problem does it solve?

The retro skill helps automate session retrospection for Claude Code agents by analyzing conversation transcripts, detecting friction, and routing findings to the appropriate destination for approval and action.

Core Features & Use Cases

  • Friction Detection: Identifies issues like tool errors, repeated mistakes, and violations of conventions.
  • Learning Capture: Captures reusable learnings and proposes action items for improvement.
  • Destination Routing: Routes findings to destinations like skill updates, project rules, or user memory for action.
  • Auto-Trigger: Optional auto-trigger for session end, enabling automated retrospection.
  • Use Case: At the end of a Claude Code session, the retro skill can automatically review the conversation transcript, detect areas for improvement, and suggest changes to enhance future sessions.

Quick Start

Use the retro skill to initiate a session retrospective with the command: /retro

Frequently Asked Questions about retro

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

FAQPage Schema
How do I automate session retrospection for Claude Code transcripts?

Automate session retrospection by analyzing Claude Code conversation transcripts to detect friction and capture learnings. The tool routes findings to appropriate destinations like skill updates or project rules for approval and action.

What is friction detection in LLM-driven agent session analysis?

Friction detection identifies issues like tool errors, repeated mistakes, and violations of conventions within session transcripts. It uses LLM classification to improve accuracy and scalability when reviewing agent conversations.

Do I need Python and jq to analyze Claude Code conversation logs?

Yes, analyzing Claude Code conversation logs requires Python3, jq, and git. You also need either gh or glab installed to route the captured findings to their appropriate destinations for action.

How do I capture reusable learnings from AI agent chat sessions?

Capture reusable learnings by reviewing chat transcripts at the end of a session. The process proposes action items for improvement and routes them to user memory, project rules, or skill updates.

Can I auto-trigger a retrospective at the end of a Claude Code session?

Yes, you can enable an optional auto-trigger for session end to perform automated retrospection. This reviews the conversation transcript, detects areas for improvement, and suggests changes to enhance future sessions.

What is the best way to update project rules based on agent conversation history?

Update project rules by routing detected frictions and captured learnings from conversation history to the project rules destination. This ensures approved changes are applied to enhance future agent sessions.