doobidoo/mcp-memory-service

Open-source persistent memory for AI agent pipelines (LangGraph, CrewAI, AutoGen) and Claude. REST API + knowledge graph + autonomous consolidation.

What it solves

AI agents and assistants typically suffer from "amnesia" between sessions, requiring users to re-explain project context, architecture decisions, and code patterns every time a new chat starts. This project provides a persistent, shared memory backend that allows agents to retrieve prior decisions and context across different runs and tools without relying on expensive cloud APIs or facing context window limits.

How it works

The service acts as a self-hosted memory layer that can be accessed via a REST API or the Model Context Protocol (MCP). It uses a hybrid search approach (BM25 and vector search) and supports multiple backends including SQLite, Milvus, and Cloudflare for synchronization. To keep data private and efficient, embeddings are run locally via ONNX. It also implements a knowledge graph with typed edges to track causal relationships (e.g., "causes", "fixes", "contradicts") rather than just storing flat facts.

Who it’s for

Developers building AI agent pipelines (using frameworks like LangGraph, CrewAI, or AutoGen) and users of AI-powered IDEs and chat interfaces (such as Claude Desktop, Cursor, and claude.ai) who need their AI to maintain long-term, cross-session memory.

Highlights

  • Multi-Transport Support: Accessible via REST API, MCP, and SSE events for real-time notifications.
  • Remote MCP: Enables native integration with the browser-based claude.ai without requiring a desktop client.
  • Local Intelligence: Runs embeddings locally using ONNX to ensure data privacy and $\text{cost-free}$ operation.
  • Knowledge Graph: Stores information as a graph with typed edges to represent causal chains.
  • Enterprise Ready: Includes OAuth 2.0 authentication, a web dashboard for memory management, and support for multi-device sync.

Related

  • Project
    Lyellr88/marm-memory

    A persistent local memory server for AI agents using the Model Context Protocol (MCP), providing shared long-term memory, code graphs, and concept graphs across sessions.

  • Project
    mem9-ai/mem9

    A persistent memory layer for AI agents that enables shared context and recall across sessions, machines, and different agent runtimes.

  • Project
    rohitg00/agentmemory

    A persistent memory server for AI coding agents that automatically captures and retrieves project context across sessions using hybrid search.

  • Project
    TencentCloud/TencentDB-Agent-Memory

    A shared memory server and management hub that allows AI agents to persist and share experience, skills, and project knowledge across different sessions and frameworks.

  • Project
    Fergana-Labs/stash

    A persistent knowledge base and shared memory layer for AI agents that connects to various data sources and records session transcripts to maintain context across runs.