OpenViSTAI · Agents · Blockchain · Crypto
Submit a project
Menu
Agent Memory & RAG

MCP Memory Service

An open-source, self-hosted memory backend for AI-agent pipelines, combining persistent storage, causal knowledge graphs and context retrieval through REST and MCP interfaces.

Explore the practical guide ↓

Independent OpenViST project profile, based on linked sources.

MCP Memory Service — OpenViST project profile in Agent Memory & RAG.
OpenViST project illustration. See the linked sources for product details.
Understand the project

Practical overview

When it fits

Use it when an AI-agent system needs persistent, shared context across sessions or agents, especially when both REST and MCP access are useful.

Who it is for

  • Developers building LangGraph, CrewAI or AutoGen agent pipelines
  • Teams connecting multiple agents or sessions to shared memory
  • Users of Claude Desktop, Claude Code or OpenCode who need persistent context
  • Engineers seeking CLI, dashboard, REST and MCP access to one memory service

Documented uses

  • Store agent decisions and project context for later sessions
  • Share memory across agents and pipeline runs
  • Retrieve context through REST endpoints or MCP
  • Represent relationships between facts in a causal knowledge graph
  • Maintain longer-term context through autonomous consolidation
Before you choose

Requirements & limits

The project is documented as open-source and provides REST and MCP access. The repository identifies an Apache-2.0 license and supports CLI, cloud, self-hosted and server deployment contexts.

  • REST and MCP transports are both documented.
  • The service includes CLI and dashboard access in addition to APIs.
  • Agent-scoped retrieval can use the X-Agent-ID header and tag search.
  • The project documents local ONNX embeddings.
  • The implementation is Python-based and licensed under Apache-2.0.

Check current requirements in the official documentation. Source-based guidance does not mean OpenViST has run or independently tested this project.

Source-based practical guide

Connect an agent pipeline through the REST API

Set up the documented HTTP service and use its memory endpoints for storage and retrieval.

Before you start

  • Python package installation with pip
  • An agent pipeline or HTTP client
  • The mcp-memory-service package

Steps

  1. Install the package with the command below.

    Documented command
    pip install mcp-memory-service

  2. Start the HTTP service with MCP_ALLOW_ANONYMOUS_ACCESS=true memory server --http.

  3. Send memory data to POST /api/memories.

  4. Search stored context with POST /api/search or retrieve by tag with POST /api/search/by-tag.

  5. Use X-Agent-ID on a store request to tag memories for agent-scoped retrieval.

Expected outcome

An HTTP-accessible memory service with documented storage and search routes for agent pipelines.

Guide limitations

  • The supplied documentation does not provide integration implementation details or test results.
  • The anonymous-access setting shown in the quick start is part of the documented setup and should be evaluated for the deployment context.
Documentation 1 ↗
Source-based practical guide

Add MCP memory access to Claude Desktop

Install the service and configure Claude Desktop to use its local MCP server.

Before you start

  • The mcp-memory-service package
  • Claude Desktop
  • Access to the Claude Desktop configuration file

Steps

  1. Install the package with the command below.

    Documented command
    pip install mcp-memory-service

  2. Add the documented memory MCP server configuration to Claude Desktop's configuration file.

  3. Use the platform-specific configuration path for macOS, Windows or Linux.

  4. Restart Claude Desktop.

Expected outcome

Claude Desktop is configured with the documented local memory MCP server.

Guide limitations

  • The guide covers configuration and restart, but does not document a verification test or further memory workflow.
Documentation 1 ↗
Evidence and transparency

Sources & review

Official repository ↗
Source record details
Source availability
Active
Sources checked
1 October 2026
Original website title
doobidoo/mcp-memory-service · Official homepage title · retained in its source language
How we review sourcesSuggest a correction