How many messages are needed for learning to trigger?

The default threshold is set to 10 messages, ensuring the skill only extracts patterns from substantial, context-rich sessions rather than short commands.

Does this slow down my Claude Code interactions?

No, it runs as a non-blocking post-session task, ensuring your active coding experience remains fast and responsive while the analysis happens in the background.

Can I control what patterns are saved?

Yes, you can customize the config.json file to define specific pattern types to detect and set ignore-patterns for items like simple typos or one-time fixes.

Where are the learned skills stored?

By default, learned skills are saved to your local filesystem at ~/.claude/skills/learned/, making them easy to manage, edit, or share.

How does the Continuous Learning skill capture information?

It uses a 'Stop' hook that triggers at the end of a session, analyzing the full transcript to identify, format, and save reusable logic as new skills.

Continuous Learning & Pattern Extraction

Name: Continuous Learning & Pattern Extraction
Author: affaan-m

byaffaan-m

•

34,605

生产力与工作流

Automatically extracts reusable coding patterns and debugging techniques from Claude Code sessions to build a personalized knowledge base.

关于

The Continuous Learning skill transforms every Claude Code interaction into a growth opportunity by automatically analyzing completed sessions for reusable insights. By leveraging non-blocking Stop hooks, it identifies successful error resolutions, framework workarounds, and project-specific conventions, saving them as permanent skills in your local configuration. This process eliminates the need to repeat manual corrections or rediscover complex fixes, allowing your AI assistant to evolve alongside your codebase and personal development style.

主要功能

Persistent Skill Storage: Saves learned patterns directly to the Claude skills directory for immediate use in future sessions.
Hook-Based Execution: Runs silently as a Stop hook to ensure 100% reliability without adding latency.
Customizable Thresholds: Configurable extraction sensitivity and ignore-lists to ensure high-quality skill generation.
Pattern Recognition: Detects error resolutions, user-provided corrections, and library workarounds.
Automatic Session Evaluation: Analyzes message history length and quality before triggering extraction.
34,605 GitHub stars

使用场景

Capturing rare framework bug workarounds that are difficult to remember or document manually.
Building a library of project-specific coding conventions and architectural patterns automatically.
Preventing repeated mistakes by automatically learning from user corrections in real-time.

关于

主要功能

Persistent Skill Storage: Saves learned patterns directly to the Claude skills directory for immediate use in future sessions.
Hook-Based Execution: Runs silently as a Stop hook to ensure 100% reliability without adding latency.
Customizable Thresholds: Configurable extraction sensitivity and ignore-lists to ensure high-quality skill generation.
Pattern Recognition: Detects error resolutions, user-provided corrections, and library workarounds.
Automatic Session Evaluation: Analyzes message history length and quality before triggering extraction.
34,605 GitHub stars

使用场景

Capturing rare framework bug workarounds that are difficult to remember or document manually.
Building a library of project-specific coding conventions and architectural patterns automatically.
Preventing repeated mistakes by automatically learning from user corrections in real-time.