Name: MCP AI Memory
Availability: InStock
Author: scanadi

What it does

MCP AI Memory connects your AI assistant to a PostgreSQL database with pgvector support, enabling it to store, retrieve, and manage contextual knowledge across sessions. It provides semantic search over stored memories using local embeddings (Transformers.js), so your agent can recall relevant information without sending data to external APIs.

Tools

memory_search — Search stored memories using natural language queries with vector similarity (default limit: 10)
memory_store — Save new information after checking for duplicates
memory_list — List all memories chronologically as a fallback when search fails
memory_update — Modify existing memory metadata
memory_delete — Remove specific memories
memory_batch — Bulk import multiple memories efficiently
memory_relate — Create relationships between memories (references, contradicts, supports, extends, causes, precedes)
memory_traverse — Traverse the memory graph using BFS/DFS algorithms with depth limits
memory_consolidate — Cluster and merge similar memories
memory_decay_status — Check lifecycle status of a memory (active, dormant, archived, expired)
memory_preserve — Protect important memories from automatic decay
memory_stats — View database statistics and queue health

Installation

Add to your claude_desktop_config.json:

{
  "mcpServers": {
    "memory": {
      "command": "npx",
      "args": ["-y", "mcp-ai-memory"],
      "env": {
        "DATABASE_URL": "postgresql://username:password@localhost:5432/memory_db"
      }
    }
  }
}

For optional Redis caching:

{
  "mcpServers": {
    "memory": {
      "command": "npx",
      "args": ["-y", "mcp-ai-memory"],
      "env": {
        "DATABASE_URL": "postgresql://username:password@localhost:5432/memory_db",
        "REDIS_URL": "redis://localhost:6379",
        "EMBEDDING_MODEL": "Xenova/all-MiniLM-L6-v2"
      }
    }
  }
}