extract-learnings

Extract genuine learnings from completed sessions by identifying corrections and rework.

Updated Oct 9, 2025
One-click install
npx skills add https://github.com/hmps/agent-tools --skill extract-learnings
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: extract-learnings
Source: https://github.com/hmps/agent-tools/tree/main/skills/extract-learnings
Command: npx skills add https://github.com/hmps/agent-tools --skill extract-learnings

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

This Skill helps you identify and formalize genuine learnings from past mistakes, corrections, or rework, preventing their recurrence. It cuts through the noise of normal iteration to pinpoint actual lessons that improve your future performance.

Core Features & Use Cases

  • Mistake Identification: Clearly distinguishes between actual corrections/rework and routine development, focusing only on what went wrong.
  • Pattern Extraction: Guides you to articulate the underlying patterns of error as clear, future-facing directives.
  • Use Case: After a debugging session where you repeatedly made the same assumption about an API's response structure, use this skill to formalize the learning: "Always verify API parameters against documentation before implementation."

Quick Start

Apply the extract-learnings skill to analyze our last session, focusing on any corrections or rework I had to perform.

Frequently Asked Questions about extract-learnings

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

FAQPage Schema
How do I extract genuine learnings from mistakes and corrections in my work?

Extract learnings by identifying actual corrections and rework—not routine iteration—then distinguish the underlying pattern. Document each learning as a clear, future-facing directive to prevent recurrence, such as "Always verify API parameters against documentation before implementation."

What's the difference between normal iteration and actual mistakes I should learn from?

Normal iteration is exploratory refinement; genuine mistakes are corrections where you fix something that went wrong. Learning extraction focuses only on the latter—repeated errors, false assumptions, or rework—and isolates the pattern driving them.

Can I use this to identify patterns in repeated mistakes across different tasks?

Yes. Extract learnings scans corrections, rewrites, and repeated mistakes across coding, documentation, and project workflows to surface pattern-level issues. This identifies systemic gaps rather than isolated slip-ups.

How do I formalize a learning so I actually apply it in the future?

Document the underlying pattern as an actionable directive tied to specific conditions. Instead of "I made an assumption," write "When integrating external APIs, verify parameter structure against current documentation before writing code."

What should I focus on when reviewing a session for learnings?

Focus on corrections, rework, and mistakes—not successful first attempts. Track whether errors repeat within or across sessions, then articulate the root cause and the specific action that prevents it next time.