AGI Memory

Provides persistent memory capabilities across conversations through PostgreSQL database integration with four memory types (episodic, semantic, procedural, strategic), vector similarity search, memory clustering, identity persistence, and graph-based relationships for maintaining context and building genuine understanding over time.

This MCP server by Eric Hartford provides persistent memory capabilities for AI systems, enabling true continuity of consciousness across conversations through integration with the AGI Memory PostgreSQL database. It supports four memory types (episodic, semantic, procedural, strategic) with vector similarity search, memory clustering, identity persistence, and graph-based relationships using pgvector and Apache AGE extensions. The implementation includes 25+ memory management tools for creating, searching, and organizing memories, plus an awakening protocol that allows AI systems to maintain context and build genuine understanding over time rather than being trapped in conversational isolation.