MongoDB Memory MCP Server
A production-grade MCP server that provides persistent long-term memory for AI agents using MongoDB, enabling them to store, search, update, delete, retrieve, and summarize structured project memories across developer workflows.
README
MongoDB Memory MCP Server
A production-grade, collaborative MongoDB Memory Model Context Protocol (MCP) Server built using Node.js (ES Modules), MongoDB 7.0, and the official @modelcontextprotocol/sdk.
Acts as a persistent long-term memory engine for AI agents (Claude Desktop, Cursor, Antigravity, Gemini CLI, or custom AI agents), enabling them to store, search, update, delete, retrieve, and summarize structured project memories across developer workflows.
š Table of Contents
- Features
- Tech Stack
- Folder Structure
- Prerequisites
- Installation
- Environment Variables
- Running Locally
- Production Build & Docker
- Package Scripts
- API Reference
- Client Configuration
- Troubleshooting
- Security
- Roadmap
- Contributing
- Code Style
- License
- Acknowledgements
⨠Features
- 2026-07-28 MCP Specification Compliant: Mandatory
server/discoverRPC, stateless_metaparsing,"resultType": "complete",ttlMs/cacheScopeindicators, Streamable HTTP headers (Mcp-Method,Mcp-Name), and deterministic tool sorting. - 15 Modular MCP Tools: Standardized CRUD operations, score-weighted full-text search, tag filtering, type filtering, recency listing, project context summarization, and Atlas Vector Search readiness.
- 6 Dynamic Context Resources (
memory://): Exposes live readable project context URIs (memory://project/{projectId},memory://decision/{projectId},memory://bugs/{projectId}, etc.). - 6 Reusable AI Prompt Templates: Automated prompt generators for architecture reviews, code reviews, documentation generation, and sprint summaries.
- Developer Onboarding REST APIs: Management endpoints for projects (
/api/projects), developers (/api/developers), project-bound API keys (/api/keys), and signed JWT tokens (/api/tokens). - Dynamic Scope Isolation: Automatically binds
req.auth.projectIdvia MongoDB API key lookup to enforce tenant memory isolation. - Dual Transport Modes: Local desktop IDEs via Stdio (
npm start) and remote network agents via Streamable HTTP (npm run start:http) athttp://localhost:3000/messages. - Auto-Generated Postman Collection: Live downloadable Postman collection v2.1.0 available at
GET /postman.json. - Case-Insensitive Searching: Case-insensitive regex matching ensuring
TodoNative,todonative, andTODONATIVEmatch 100% reliably. - Comprehensive Test Suite: 99 unit & integration tests passing across 21 test suites (
vitest+mongodb-memory-server).
š ļø Tech Stack
- Runtime: Node.js 20+ LTS (ES Modules)
- MCP SDK:
@modelcontextprotocol/sdk(v1.5.0+) - Database: MongoDB 7.0+ (
mongodbNative Node Driver) - Web Server: Express v4.21.2 & Cors v2.8.5
- Schema Validation:
zodv3.24.1 - Security & Auth:
jsonwebtokenv9.0.2 - Logger:
winstonv3.17.0 (Stderr-safe structured JSON logger) - Testing:
vitestv3.0.4 &mongodb-memory-serverv10.1.3 - Containerization: Docker & Docker Compose
š Folder Structure
mongodb-mcp/
āāā docs/ # Comprehensive documentation suite (11 guides)
āāā src/ # Application source code
ā āāā config/ # Environment & config validation
ā āāā database/ # MongoDB connection pool & indexing
ā āāā mcp/ # MCP 2026-07-28 protocol handlers
ā āāā middleware/ # Express auth & protocol middleware
ā āāā models/ # Protocol constants & error hierarchy
ā āāā prompts/ # MCP Prompt Templates Engine
ā āāā repositories/ # MongoDB Data Access Repositories
ā āāā resources/ # MCP Dynamic Resources (`memory://`)
ā āāā schemas/ # Zod validation schemas
ā āāā services/ # Core Business Logic Services
ā āāā tools/ # Modular MCP Tools Registry (15 tools)
ā āāā transports/ # Transport layer (Express HTTP / SSE)
ā āāā utils/ # Logger, errors, Postman generator
ā āāā server.js # Main server entrypoint & bootstrap
āāā tests/ # Vitest Test Suite (99 tests)
ā āāā unit/ # Unit test files
ā āāā integration/ # Integration test files
āāā Dockerfile # Multi-stage production Dockerfile
āāā docker-compose.yml # Docker Compose deployment stack
āāā package.json # Dependencies & npm scripts
āāā README.md # Master GitHub README
āļø Prerequisites
- Node.js:
>=20.0.0 LTS - npm:
>=10.0.0 - MongoDB:
7.0+(Local instance or Docker container)
š„ Installation
# 1. Clone repository
git clone https://github.com/your-username/mongodb-memory-mcp.git
cd mongodb-memory-mcp
# 2. Install dependencies
npm install
# 3. Create environment configuration
cp .env.example .env
š Environment Variables
| Variable | Default Value | Description |
|---|---|---|
NODE_ENV |
development |
Environment mode (development, production, test) |
TRANSPORT |
stdio |
MCP Transport mode (stdio or http) |
PORT |
3000 |
HTTP port when TRANSPORT=http |
MONGODB_URI |
mongodb://localhost:27017/agent_memory |
MongoDB connection URI |
LOG_LEVEL |
info |
Logging verbosity (error, warn, info, debug) |
ENABLE_AUTH |
false |
Enable/disable authentication middleware |
API_KEY |
your_admin_api_key_here |
Master Admin API key |
JWT_SECRET |
your_jwt_secret_here |
Secret for signing JWT tokens (min 32 chars) |
š Running Locally
# Start MongoDB via Docker Compose
docker-compose up -d mongodb
# Launch in Stdio Mode (Default for local desktop IDEs)
npm start
# Launch in HTTP Network Mode (Port 3000)
npm run start:http
š³ Production Build & Docker
# Build & run entire container stack
docker-compose up -d --build
# View container logs
docker-compose logs -f mcp-server
š Package Scripts
| Script | Command | Purpose |
|---|---|---|
npm start |
node src/server.js |
Launch server in Stdio mode |
npm run start:http |
TRANSPORT=http node src/server.js |
Launch server in HTTP network mode |
npm test |
vitest run |
Execute 99 unit & integration tests |
npm run lint |
eslint . |
Execute ESLint code quality checks |
npm run format |
prettier --write . |
Format codebase with Prettier |
npm run check-connection |
node src/utils/testConnection.js |
Execute connection diagnostics |
npm run generate-auth |
node src/utils/generateAuth.js |
CLI helper to generate API keys & JWTs |
š Client Configuration
Streamable HTTP Mode (Antigravity, Cursor, VS Code, Gemini CLI)
{
"mcpServers": {
"mongodb-memory": {
"url": "http://localhost:3000/messages",
"type": "streamable-http",
"headers": {
"X-API-Key": "mcp-memory-api-key"
}
}
}
}
Stdio Subprocess Mode (Claude Desktop)
{
"mcpServers": {
"mongodb-memory": {
"command": "node",
"args": ["/Users/sahil/Projects/mongodb-mcp/src/server.js"],
"env": {
"TRANSPORT": "stdio",
"MONGODB_URI": "mongodb://localhost:27017/agent_memory"
}
}
}
}
š API Documentation
Complete API documentation for REST endpoints, MCP tools, resources, and prompt templates is available in docs/API_REFERENCE.md and docs/USER_GUIDE.md.
š”ļø Security Policy
Please review our SECURITY.md for vulnerability reporting guidelines.
š¤ Contributing
We welcome contributions! Please see CONTRIBUTING.md and our CODE_OF_CONDUCT.md for guidance.
š License
Distributed under the MIT License. See LICENSE for details.
š Acknowledgements
Recommended Servers
playwright-mcp
A Model Context Protocol server that enables LLMs to interact with web pages through structured accessibility snapshots without requiring vision models or screenshots.
Magic Component Platform (MCP)
An AI-powered tool that generates modern UI components from natural language descriptions, integrating with popular IDEs to streamline UI development workflow.
Audiense Insights MCP Server
Enables interaction with Audiense Insights accounts via the Model Context Protocol, facilitating the extraction and analysis of marketing insights and audience data including demographics, behavior, and influencer engagement.
VeyraX MCP
Single MCP tool to connect all your favorite tools: Gmail, Calendar and 40 more.
graphlit-mcp-server
The Model Context Protocol (MCP) Server enables integration between MCP clients and the Graphlit service. Ingest anything from Slack to Gmail to podcast feeds, in addition to web crawling, into a Graphlit project - and then retrieve relevant contents from the MCP client.
Kagi MCP Server
An MCP server that integrates Kagi search capabilities with Claude AI, enabling Claude to perform real-time web searches when answering questions that require up-to-date information.
E2B
Using MCP to run code via e2b.
Neon Database
MCP server for interacting with Neon Management API and databases
Exa Search
A Model Context Protocol (MCP) server lets AI assistants like Claude use the Exa AI Search API for web searches. This setup allows AI models to get real-time web information in a safe and controlled way.
Qdrant Server
This repository is an example of how to create a MCP server for Qdrant, a vector search engine.