Where are my learned skills stored?

Extracted patterns are saved to ~/.claude/skills/learned/, allowing you to review, edit, or share the skills that Claude has developed from your sessions.

How does the Continuous Learning skill work?

It functions as a 'Stop' hook that triggers at the end of a Claude Code session. It evaluates the conversation history for repeatable patterns and saves them as 'learned skills' in your local configuration directory.

Can I choose what Claude learns?

Yes. Through the config.json file, you can define specific patterns to detect, set minimum session lengths for evaluation, and establish ignore lists for trivial fixes or typos.

Will this impact my coding performance?

No. Because it runs as a stop hook after you have finished your session, it does not add any latency to your active chat or tool-use commands.

Continuous Learning Agent

Name: Continuous Learning Agent
Author: unju-ai

byunju-ai

0•

Productivity & Workflow

Automates the extraction of reusable coding patterns and session insights to build a persistent, personalized knowledge base for Claude Code.

The Continuous Learning skill transforms Claude Code from a transient assistant into an evolving partner by capturing domain-specific knowledge during every interaction. Operating as a non-blocking 'Stop' hook, it analyzes completed sessions to identify successful error resolutions, project-specific conventions, and unique library workarounds. By distilling these into permanent skills saved locally, it ensures that once a problem is solved or a pattern is established, the AI retains that wisdom for all future sessions, effectively creating a self-improving development environment tailored to your specific codebase.

Key Features

01Seamless integration with existing Claude settings and commands

020 GitHub stars

03Local skill storage for persistent long-term memory

04Pattern recognition for error resolutions and debugging workflows

05Automated session evaluation via non-blocking Stop hooks

06Customizable extraction thresholds and ignore lists

Use Cases

01Capturing and reusing complex workarounds for specific library versions

02Automatically documenting project-specific architectural patterns and styles

03Building a personalized library of successful debugging techniques

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add unju-ai/ecc continuous-learning

For use in Claude.ai and ChatGPT