Ultimate MCP Coding Platform
Ultimate MCP is a production-ready Model Context Protocol platform that turns any LLM into a coding co-pilot. It ships with a FastAPI + FastMCP backend, Neo4j graph persistence, OpenAI Agent integration, a React frontend, and Docker Compose orchestration.
Features
Real MCP server with lint, test, execution, generation, and graph tools
Neo4j persistence for tool artefacts with aggregation metrics
REST API mirroring MCP tools and secured by bearer token + rate limiting
Structured logging, strict CORS, security headers, and per-request IDs
React + Vite frontend for human operators
OpenAI Agents SDK bridge for autonomous tool discovery and execution
Complete CI pipeline (lint, type-check, tests with coverage, Docker builds)
Docker Compose for one-command local deployment
Repository Layout
Quickstart
1. Dependencies
Python 3.13+
Node.js 20+
Docker & Docker Compose (for containerised runs)
2. Bootstrap Environment
This creates backend/.venv
, installs Python requirements, and runs npm install
for the frontend.
3. Run Locally (Developer Mode)
Backend:
Frontend:
Open the UI at http://localhost:3000
. The API docs live at http://localhost:8000/docs
.
4. Run with Docker Compose
Expose:
Frontend:
http://localhost:3000
API:
http://localhost:8000
Neo4j Browser:
http://localhost:7474
Testing
A ready-made smoke test hits key endpoints:
MCP & Agent Integration
MCP server mounted at
/mcp
using FastMCP streamable HTTP transport.backend/agent_integration/client.py
providesAgentDiscovery
for listing/invoking tools and anOpenAIAgentBridge
to register the server with OpenAI Agents.
Example usage:
Security Highlights
Bearer token auth on all mutating endpoints
SlowAPI rate limiting (default 10 req/s per IP)
Request size checks and security headers (
CSP
,X-Frame-Options
, etc.)Non-root Docker images with capabilities dropped
Detailed guidance in docs/SECURITY.md
.
Configuration
See .env.example
for required variables:
Documentation
Release Packaging
Create an archive for distribution:
License
MIT License © 2025 Ultimate MCP maintainers.
Ultimate_MCP
This server cannot be installed
hybrid server
The server is able to function both locally and remotely, depending on the configuration or use case.
Turns any LLM into a coding co-pilot with production-ready MCP server providing lint, test, execution, generation, and graph tools. Features Neo4j persistence, OpenAI Agent integration, REST API, and React frontend for comprehensive code development assistance.