continuous-learning

Extract reusable patterns from Claude Code sessions into learned skills.

Updated Nov 21, 2025

Search Tags:#claude-code #pattern-extraction #learned-skills #continuous-learning #session-end

Ready to run

npx skills add https://github.com/MBarry01/dousell-immo --skill continuous-learning-mbarry01

💬 Prompt for ChatGPT / Claude Chat▼

Please help me install this Agent Skill.
Skill: continuous-learning
Source: https://github.com/MBarry01/dousell-immo/tree/main/.claude/skills/continuous-learning
Command: npx skills add https://github.com/MBarry01/dousell-immo --skill continuous-learning-mbarry01

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Automates turning Claude Code sessions into a library of reusable skills by extracting recurring patterns.

Core Features & Use Cases

End-of-session learning: captures patterns when a session ends and saves them for future use.
Pattern taxonomy: classifies patterns into error_resolution, user_corrections, workarounds, debugging_techniques, and project_specific.
Centralized learning store: writes learned skills to .claude/skills/learned/ for quick retrieval in future sessions.
Use Case: Data scientists or developers running Claude Code can gradually accumulate actionable techniques and best practices without manual note-taking.

Quick Start

Configure the Stop hook in .claude/settings.json and run Claude Code sessions to start collecting learned patterns automatically.

Frequently Asked Questions about continuous-learning

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

FAQPage Schema

How do I extract reusable patterns from Claude Code sessions automatically?▼

You can extract reusable patterns by configuring a Stop hook in .claude/settings.json that triggers at session end, automatically detecting and saving workarounds and debugging techniques to a learned skills directory.

What types of patterns can be captured when Claude Code sessions end?▼

Session end extraction classifies patterns into five categories: error_resolution, user_corrections, workarounds, debugging_techniques, and project_specific patterns for future reuse.

How do I configure continuous learning for Claude Code sessions?▼

Continuous learning is configured via a config.json file where you set min_session_length, extraction_threshold, auto_approve, learned_skills_path, and specify which pattern categories to detect.

Do I need to manually approve every learned skill extracted from Claude Code?▼

No, manual approval is optional. You can set the auto_approve parameter to true in config.json to automatically accept extracted patterns into your learned skills store.

Where are extracted patterns stored for reuse across future Claude Code sessions?▼

Extracted patterns are saved to the .claude/skills/learned/ directory, creating a centralized learning store accessible for quick retrieval in future Claude Code sessions.

Can I filter Claude Code session learning to ignore short interactions?▼

Yes, you can filter short interactions by setting the min_session_length and extraction_threshold parameters in config.json to control when pattern extraction triggers.