M.I.M.I.R

v1.0.0Knowledge & Memorystable

Mimir - Fully open and customizable memory bank with semantic vector search capabilities for locally indexed files (Code Intelligence) and stored memories that are shared across sessions and chat contexts allowing worker agent to learn from errors in

code-intelligencecodebasedockergraph-algorithmsgraph-api
Share:
274
Stars
0
Downloads
0
Weekly
0/5

What is M.I.M.I.R?

M.I.M.I.R is a Model Context Protocol (MCP) server that allows AI assistants like Claude, Cursor, and VS Code to mimir - fully open and customizable memory bank with semantic vector search capabilities for locally indexed files (code intelligence) and stored memories that are shared across sessions and chat cont...

Mimir - Fully open and customizable memory bank with semantic vector search capabilities for locally indexed files (Code Intelligence) and stored memories that are shared across sessions and chat contexts allowing worker agent to learn from errors in

This server falls under the Knowledge & Memory category on MCPgee, the world's largest MCP server directory with 33,000+ servers.

Features

  • Mimir - Fully open and customizable memory bank with semanti

Use Cases

Create semantic vector search over codebases
Store shared memories across sessions
Enable code intelligence for agents
orneryd

Maintainer

LicenseMIT License
Languagego
Versionv1.0.0
UpdatedMay 21, 2026
Statushealthy
Maintenanceactive

Works with

ClaudeOpenAIwindowsmacoslinux

Installation

Manual Installation

npx m-i-m-i-r-multi-agent-intelligent-memory-insight-repository

Configuration

Configuration Details

Config File

claude_desktop_config.json

Performance

Response Metrics

Response Time< 200ms
ThroughputMedium

Resource Usage

Memory UsageLow
CPU UsageLow

How to Set Up and Use M.I.M.I.R

M.I.M.I.R (Multi-agent Intelligent Memory and Insight Repository) is a fully open, customizable memory bank for AI agents that combines semantic vector search with a Neo4j graph database to index local codebases and store persistent memories shared across sessions. It exposes 13 MCP tools covering memory creation, relationship management, semantic search, file indexing, and task management, enabling worker agents to learn from past errors and retrieve relevant context without re-reading entire codebases. Go-powered and Docker-based, it runs locally with support for multiple LLM providers including OpenAI, GitHub Copilot, and Ollama.

Prerequisites

  • Docker and Docker Compose installed (the server runs as a multi-container stack)
  • Git to clone the repository
  • An LLM provider API key or a running Ollama instance (for OpenAI: MIMIR_LLM_API_KEY; for Copilot or Ollama: endpoint URL)
  • At least 4 GB RAM recommended for Neo4j and the embedding model
  • An MCP-compatible client such as Claude Desktop or Cursor
1

Clone the repository

Clone the Mimir repository to your local machine.

git clone https://github.com/orneryd/Mimir.git
cd Mimir
2

Configure environment variables

Copy the example environment file and edit it with your LLM provider credentials and workspace path.

cp env.example .env
# Edit .env with your settings:
# HOST_WORKSPACE_ROOT=~/src
# NEO4J_PASSWORD=yourpassword
# MIMIR_DEFAULT_PROVIDER=openai
# MIMIR_LLM_API_KEY=sk-your-key
# MIMIR_DEFAULT_MODEL=gpt-4o
3

Start the full stack

Run the startup script which automatically detects your platform and launches Neo4j, the embedding service, and the Mimir MCP server.

npm run start
4

Verify services are running

Check that the web UI, Neo4j browser, and MCP API are accessible.

curl http://localhost:9042/health
# Web UI: http://localhost:9042
# Neo4j Browser: http://localhost:7474
# MCP endpoint: http://localhost:9042/mcp
5

Configure your MCP client

Add Mimir to your MCP client configuration pointing at the local MCP endpoint.

{
  "mcpServers": {
    "mimir": {
      "command": "npx",
      "args": ["m-i-m-i-r-multi-agent-intelligent-memory-insight-repository"],
      "env": {
        "MIMIR_MCP_URL": "http://localhost:9042/mcp"
      }
    }
  }
}
6

Index your codebase

Use the index_folder tool to add your source code directory to Mimir's semantic index. Set HOST_WORKSPACE_ROOT in your .env to the parent directory of your projects.

# In your .env:
HOST_WORKSPACE_ROOT=/Users/you/src
MIMIR_EMBEDDINGS_ENABLED=true
MIMIR_EMBEDDINGS_MODEL=bge-m3

M.I.M.I.R Examples

Client configuration

MCP client configuration for a locally running Mimir instance.

{
  "mcpServers": {
    "mimir": {
      "command": "npx",
      "args": ["m-i-m-i-r-multi-agent-intelligent-memory-insight-repository"],
      "env": {
        "MIMIR_MCP_URL": "http://localhost:9042/mcp",
        "NEO4J_PASSWORD": "yourpassword"
      }
    }
  }
}

Prompts to try

Example prompts for working with Mimir's memory and code intelligence tools.

- "Search my codebase for functions that handle authentication"
- "Remember that the payments module uses idempotency keys to prevent duplicate charges"
- "Create a task: refactor the user service to use dependency injection, priority high"
- "What did we learn from the last deployment failure? Search memories about errors"
- "Show all indexed directories and their embedding statistics"

Troubleshooting M.I.M.I.R

Neo4j fails to start or memory errors during startup

Increase Docker's memory allocation to at least 4 GB in Docker Desktop settings. Neo4j requires significant heap space, especially with large codebases indexed. Also check that NEO4J_PASSWORD in .env matches what Docker Compose expects.

Semantic search returns no results after indexing

Ensure MIMIR_EMBEDDINGS_ENABLED=true in your .env and that the embeddings model (bge-m3 or nomic-embed-text) has finished downloading. Check embedding statistics via the MCP tool to confirm vectors were generated.

MCP client cannot connect to http://localhost:9042/mcp

Verify the stack is running with `docker compose ps`. If the mimir service shows as unhealthy, check logs with `docker compose logs mimir`. Ensure port 9042 is not blocked by a firewall or used by another process.

Frequently Asked Questions about M.I.M.I.R

What is M.I.M.I.R?

M.I.M.I.R is a Model Context Protocol (MCP) server that mimir - fully open and customizable memory bank with semantic vector search capabilities for locally indexed files (code intelligence) and stored memories that are shared across sessions and chat contexts allowing worker agent to learn from errors in It connects AI assistants to external tools and data sources through a standardized interface.

How do I install M.I.M.I.R?

Follow the installation instructions on the M.I.M.I.R GitHub repository. Clone the repo, install dependencies, and add the server config to your AI client.

Which AI clients work with M.I.M.I.R?

M.I.M.I.R works with all major MCP-compatible AI clients including Claude Desktop, Claude Code, Cursor, VS Code (GitHub Copilot), Windsurf, and Cline.

Is M.I.M.I.R free to use?

Yes, M.I.M.I.R is open source and available under the MIT License license. You can use it freely in both personal and commercial projects.

Browse More Knowledge & Memory MCP Servers

Explore all knowledge & memory servers available in the MCPgee directory. Each server includes setup guides for Claude, Cursor, and VS Code.

Quick Config Preview

{ "mcpServers": { "m-i-m-i-r-multi-agent-intelligent-memory-insight-repository": { "command": "npx", "args": ["-y", "m-i-m-i-r-multi-agent-intelligent-memory-insight-repository"] } } }

Add this to your claude_desktop_config.json or .cursor/mcp.json

Read the full setup guide →

Ready to use M.I.M.I.R?

Browse our complete directory of 33,000+ MCP servers, read setup guides for your editor, and start building with the Model Context Protocol.

33,000+ ServersFree & Open SourceStep-by-Step Guides