The Procedural Memory System (PMS) implements a sophisticated three-tier memory architecture—Episodic, Semantic, and Procedural—designed to transform raw conversation transcripts into actionable, long-term knowledge. By analyzing recurring coding patterns, user preferences, and decision-making workflows across multiple sessions, PMS automatically synthesizes context-aware rules that persist in Claude's environment. This skill is essential for developers who want Claude to learn their specific architectural styles, anti-patterns, and project-specific requirements over time, ensuring a more personalized and efficient AI pairing experience.
Key Features
01Automatic session transcript encoding with built-in privacy redaction
020 GitHub stars
03Semantic pattern extraction for user preferences and code styles
04Configurable sensitivity thresholds and automatic event hooks
05Three-tier memory architecture for long-term knowledge retention
06Automated synthesis of persistent rules in the .claude/rules/ directory
