Captures and surfaces tacit knowledge from coding sessions to provide persistent context for AI assistants.
Working Memory (wm) acts as a persistent memory layer for AI coding assistants, automatically capturing the unspoken rationale, patterns, and constraints discovered during development. By distilling session-specific wisdom into structured files, it ensures that your AI assistant learns how you work and retains critical project context across multiple interactions, preventing redundant explanations and improving the relevance of AI-generated suggestions.
主要功能
01Session management to archive and review development history
02Dive-prep agent for deep context gathering and project grounding
03Memory compression to synthesize high-level insights from long sessions
043 GitHub stars
05Automatic capture of tacit knowledge and architectural rationale
06Distillation of wisdom patterns (Metis) and project guardrails
使用场景
01Onboarding AI assistants to complex codebases with unique constraints
02Reducing token usage by synthesizing redundant session history into high-level state
03Maintaining consistent architectural patterns across long-running development projects
