ChromaDB Persistent Knowledge & Recall System

Achieving Continuous Context Awareness

Every new chat with an AI starts with amnesia. You constantly say "I prefer Tailwind CSS" or "Don't use Python classes for this". The Long-term Memory MCP solves this. It maintains a running index in a local vector database (like Chroma or Qdrant). When you ask a question, the AI queries its own memory first for relevant hidden context before answering.

Core Use Cases

1. "Remember that my primary email address for deployments is admin@domain.com."
2. "Last week we discussed a custom Python decorator for caching. I forgot the logic, fetch it from memory and apply it to this logic block."

Server Configuration Example

"mcpServers": {
  "memory": {
    "command": "python3",
    "args": ["-m", "rag_memory_mcp"]
  }
}