reflection

Extract actionable lessons from a session and update CLAUDE.md.

109|17|Updated May 8, 2021
One-click install
npx skills add https://github.com/straubt1/tfx --skill reflection-straubt1
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: reflection
Source: https://github.com/straubt1/tfx/tree/main/.claude/skills/reflection
Command: npx skills add https://github.com/straubt1/tfx --skill reflection-straubt1

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Collates and preserves actionable lessons from a user session to prevent repeated mistakes and to improve future interactions.

Core Features & Use Cases

  • Session retrospective: Analyzes the entire conversation to surface decisions, corrections, and preferences.
  • CLAUDE.md guidance: Proposes precise updates to CLAUDE.md to encode lessons for future sessions.
  • Structured insights: Outputs categorized lessons (patterns, pitfalls, architecture, workflow) for easy review.

Quick Start

Run reflection on the current session to extract and persist lessons to CLAUDE.md.

Frequently Asked Questions about reflection

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

FAQPage Schema
How do I capture lessons learned from a Claude session to improve future interactions?

To capture lessons learned, a session retrospective analyzes your full conversation history to surface decisions, corrections, and preferences, then proposes structured updates to your CLAUDE.md file to encode those lessons for future iterations.

How does a session retrospective filter conversation noise when generating documentation?

A session retrospective reviews the entire conversation history and filters out noise to identify actionable insights, ensuring safety and accuracy by avoiding sensitive data leakage while extracting categorized patterns, pitfalls, architecture, and workflow notes.

What is the best way to update CLAUDE.md with structured notes from a previous session?

The best way to update CLAUDE.md is to run a session review that proposes precise, structured edits encoding categorized lessons such as patterns and pitfalls, ensuring the documentation remains searchable and useful for future interactions.

Can I automatically generate a retrospective from my conversation history without manual review?

Yes, a retrospective can automatically analyze your conversation history to extract decisions and corrections, categorizing them into structured insights like architecture and workflow notes while filtering noise and preventing sensitive data leakage.

What limitations exist when extracting actionable lessons from session documentation?

Limitations include the need to filter conversational noise and actively avoid sensitive data leakage, meaning the extracted lessons require structured categorization into patterns, pitfalls, architecture, and workflow to remain accurate and actionable.