Captures and persists user corrections and preferences into CLAUDE.md files to enable continuous AI learning.
Claude Reflect is a specialized self-learning system that enables Claude Code to remember your specific preferences and corrections across sessions. By monitoring for natural correction patterns such as 'actually' or 'use X instead of Y,' the skill automatically queues insights that can be reviewed and synced to your project or global CLAUDE.md files. This creates a persistent feedback loop that reduces repetitive instructions and ensures the AI adheres to your unique coding standards, architectural patterns, and tool preferences automatically.
Key Features
01Queued learning system for human-in-the-loop review before file updates
02Support for both project-specific and global CLAUDE.md destinations
03Dry-run mode to preview documentation changes without applying them
040 GitHub stars
05Automatic detection of correction patterns like 'no, use X' or 'remember this'
06History scanning to recover missed learnings from previous sessions
Use Cases
01Enforcing specific library versions or API preferences across a project
02Automating the documentation of project-specific naming conventions
03Building a global user profile of general AI behavior preferences
