Nife.io MCP Server

A Model Context Protocol (MCP) server that interfaces with the Nife.io GraphQL API

Installation • Quick Start • Documentation • Contributing

Overview

Nife MCP Server provides a standardized interface to interact with Nife.io's GraphQL API through the Model Context Protocol, enabling seamless integration with Claude Desktop and other MCP-compatible applications.

Key Features

• Intelligent Schema Discovery - Automatically discovers and adapts to the GraphQL schema

• Dynamic Tool Generation - Creates MCP tools on-the-fly from schema

• Zero Configuration - Works out of the box with minimal setup

• Secure Authentication - Bearer token support with flexible configuration

• Multiple Install Methods - PyPI, NPM, or GitHub

• Cross-Platform - Works on macOS, Linux, and Windows

• Production Ready - Comprehensive error handling and logging

Related MCP server: h3-mcp-tools

Installation

Option 1: PyPI (Recommended for Python users)

pip install nife-mcp-server

Option 2: NPM (Best for Claude Desktop)

npx @nifelabs/mcp-server@latest

Option 3: From Source

git clone https://github.com/nifetency/nife-mcp-server
cd nife-mcp-server
pip install -r requirements.txt

Quick Start

1. Get Your Access Token

nifectl auth login
nifectl auth token

2. Configure Claude Desktop

You can use the Nife MCP Server in three ways depending on your setup.

1. Recommended: Using NPX (No Installation Required)

This is the easiest way to get started.

{
  "mcpServers": {
    "nife": {
      "command": "npx",
      "args": ["--yes", "@nifelabs/mcp-server@latest"],
      "env": {
        "NIFE_ACCESS_TOKEN": "your_token_here",
        "PYTHON": "PYTHON_PATH"

      }
    }
  }
}

2. Using Python (Installed via pip)

If you installed the package via pip:

pip install nife-mcp-server

Use this configuration:

{
  "mcpServers": {
    "nife": {
      "command": "python",
      "args": ["-m", "nife_mcp_server"],
      "env": {
        "NIFE_ACCESS_TOKEN": "your_token_here"
      }
    }
  }
}

3. Local Development (Source Code)

If you're running from source:

{
  "mcpServers": {
    "nife": {
      "command": "python",
      "args": ["-m", "nife_mcp_server.intelligent_main"],
      "env": {
        "PYTHONPATH": "C:\\path\\to\\nife-mcp-server\\src",
        "NIFE_ACCESS_TOKEN": "your_token_here"
      }
    }
  }
}

Replace the PYTHONPATH with your local project path.

Environment Variables

Variable Description

NIFE_ACCESS_TOKEN Your API access token from Nife

3. Restart Claude Desktop

Your Nife.io integration is now active! 🎉

📚 Documentation

• Quick Start - Get started in 5 minutes

• Hosted connector (URL-based) - Deploy the remote server and connect by URL

• API Documentation - Complete API reference

• Release Guide - For contributors and maintainers

• Contributing - How to contribute

Hosted connector tools: reliability, cost and platform

The hosted connector (nife-mcp-http) adds curated tools on top of the site, app and custom-domain tools. The stdio server (intelligent_main.py) loads the same tools next to its schema-generated ones when the optional dependencies are installed (pip install "nife-mcp-server[http]", or pip install -r requirements.txt). It uses NIFE_ACCESS_TOKEN as the caller's token. Without those packages it runs with only the schema-generated tools and logs a hint on stderr. Every call is made with the caller's own token. Organization-keyed tools check membership first, and alert, billing cycle and recommendation ids are checked against the caller's own lists before use.

Reliability (NERO)

Cost (FinOps)

Platform, delivery and compliance (read-only)

These tools only read. Cloud credentials and kubeconfig locations are never requested.

Infrastructure and governance (read-only)

The forecast service address can be changed with NIFE_LSTM_URL.

Virtual machines and cloud accounts (changes)

Cron jobs

Usage

Starting the Server

# If installed via pip
nife-mcp-server

# If running from source
python -m nife_mcp_server.intelligent_main

The server speaks MCP over stdin/stdout, so it is normally launched by your MCP client (e.g. Claude Desktop) rather than run by hand. It does not listen on a network port.

To run the hosted HTTP connector instead, install the http extra (pip install "nife-mcp-server[http]") and run nife-mcp-http. It listens on PORT (default 8000); see HOSTED.md.

Environment Variables

• NIFE_ACCESS_TOKEN - Your Nife.io access token

The GraphQL endpoint is fixed to https://api.nife.io/graphql. The hosted connector has its own settings, listed in HOSTED.md.

Example Usage with Claude

Once configured, you can ask Claude:

"List all my applications in Nife.io"
"Show me the status of my deployments"
"Create a new application called 'my-app'"

Architecture

┌─────────────────┐
│  Claude Desktop │
└────────┬────────┘
         │ MCP Protocol
┌────────▼────────┐
│  Nife MCP Server│
└────────┬────────┘
         │ GraphQL
┌────────▼────────┐
│   Nife.io API   │
└─────────────────┘

Core Components

• Schema Manager - Intelligent GraphQL schema introspection

• Tool Generator - Dynamic MCP tool creation

• Query Builder - Smart query construction

• Authentication - Secure token management

API Endpoints

MCP Endpoints

GraphQL Integration

The server provides full access to Nife.io's GraphQL API:

• Query all resources (apps, deployments, etc.)

• Execute mutations (create, update, delete)

• Custom query execution

• Schema introspection

Development

Setup Development Environment

# Clone the repository
git clone https://github.com/nifetency/nife-mcp-server
cd nife-mcp-server

# Create virtual environment
python -m venv .venv
source .venv/bin/activate  # On Windows: .venv\Scripts\activate

# Install dependencies
pip install -r requirements.txt
pip install -e .  # Install in editable mode



# Run the server
cd src
python -m nife_mcp_server.intelligent_main

Project Structure

nife-mcp-server/
├── src/
│   └── nife_mcp_server/
│       ├── intelligent_main.py  # Main server logic
│       ├── schema_manager.py    # GraphQL schema handling
│       ├── http_app.py          # Hosted (Streamable HTTP) connector
│       ├── http_auth.py         # Auth for the hosted connector
│       ├── oauth.py             # OAuth support
│       ├── nife_api.py          # Async client (GraphQL, express-backend, FinOps)
│       ├── tool_utils.py        # Shared tool helpers and ownership checks
│       ├── site_tools.py        # Site deploy, manage and inspect tools
│       ├── app_tools.py         # Container app tools
│       ├── domain_tools.py      # Custom-domain tools
│       ├── ops_tools.py         # NERO and FinOps tools
│       ├── platform_tools.py    # Delivery, compliance, usage and capacity tools
│       ├── infra_tools.py       # Cloud accounts, clusters, agents, governance
│       ├── vm_tools.py          # VM power actions and cloud account checks
│       ├── cron_tools.py        # Cron job runs
│       └── routes/
│           └── mcp.py          # MCP route handlers
├── tests/                      # Test suite
├── Dockerfile                  # Container image for the hosted server
└── bin/                        # Executable scripts

Running Tests

pytest
pytest --cov=nife_mcp_server  # With coverage

Contributing

We love contributions! Please see our Contributing Guide for details.

Ways to Contribute

• Report bugs

• Suggest features

• Improve documentation

• Submit pull requests

Development Process

1. Fork the repository

2. Create a feature branch

3. Make your changes

4. Add tests

5. Submit a pull request

Stats

License

This project is licensed under the MIT License - see the LICENSE file for details.

Acknowledgments

• Built for the Model Context Protocol

• Powered by Nife.io

• Designed for Claude Desktop

Support

• Email: support@nife.io

• Issues: GitHub Issues

• Discussions: GitHub Discussions

• Documentation: Full Docs

Roadmap

• WebSocket support for real-time updates

• GraphQL subscriptions

• Enhanced caching layer

• Query batching

• Metrics and monitoring

• Docker Compose examples

• Kubernetes manifests

Star History

If you find this project useful, please consider giving it a star!

Made by the Nife.io team

Website • Documentation • Blog • MCP Open Source