retrospect

Analyze Claude Code sessions against CLAUDE.md rules and produce a structured Retrospect Report.

Updated Jan 6, 2026
One-click install
npx skills add https://github.com/devseunggwan/praxis --skill retrospect-devseunggwan
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: retrospect
Source: https://github.com/devseunggwan/praxis/tree/main/skills/retrospect
Command: npx skills add https://github.com/devseunggwan/praxis --skill retrospect-devseunggwan

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Session retrospection is a structured approach to analyzing Claude Code sessions against CLAUDE.md rules, surfacing friction patterns and root causes, and proposing context-appropriate improvements to be executed after user approval.

Core Features & Use Cases

  • Systematic capture of friction events at the end of a session and clustering into actionable insights
  • Root-cause analysis with traceability to CLAUDE.md rules and documented evidence
  • Generation of artifacts (MEMORY.md entries, GitHub issues, CLAUDE.md draft proposals, upstream feedback) with verification steps
  • A deterministic 4-stage process: Load → Analyze → Report/Approve → Execute to drive continuous improvement

Quick Start

Run the retrospective workflow at the end of a session to capture friction events, analyze them with tracer and analyst agents, and review the proposed actions in the final report for approval.

Frequently Asked Questions about retrospect

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

FAQPage Schema
How do I analyze Claude Code session friction against CLAUDE.md rules?

Session friction analysis evaluates your Claude Code sessions against CLAUDE.md rules to identify root causes and repeat patterns. It uses a deterministic 4-stage process to cluster friction events into structured retrospectives with actionable recommendations.

What is the best way to automate end-of-session retrospectives for workflow improvement?

Automated end-of-session retrospectives capture friction events, analyze them with tracer and analyst agents, and generate a structured Retrospect Report. This process surfaces repeat patterns to drive continuous workflow improvement without manual review.

How do I generate GitHub issues and MEMORY.md entries from root-cause analysis?

Root-cause analysis generates artifacts like MEMORY.md entries, GitHub issues, and CLAUDE.md draft proposals with verification steps. These artifacts are produced after the analysis stage and require user approval before execution.

Can I use session retrospection to update CLAUDE.md rules based on detected friction?

Yes, session retrospection produces CLAUDE.md draft proposals as actionable artifacts based on detected friction. The workflow integrates tracer and analyst agents to ensure traceability from friction events to documented rule improvements.

Do I need any external dependencies to run a Claude Code session retrospective?

No external dependencies are required to run a Claude Code session retrospective. The workflow operates independently through its 4-stage process of loading, analyzing, reporting, and executing friction analysis against your existing CLAUDE.md configuration.