mcp-server-prometheus

MCP server for interacting with Prometheus metrics and data.

This is a TypeScript-based MCP server that implements a Prometheus API interface. It provides a bridge between Claude and your Prometheus server through the Model Context Protocol (MCP).

Demo

demo

Features

Resources

List and access Prometheus metric schema
Each metric resource provides:
- Metric name and description
- Detailed metadata from Prometheus
- Statistical information (count, min, max)
JSON mime type for structured data access

Current Capabilities

List all available Prometheus metrics with descriptions
Read detailed metric information including:
- Metadata and help text
- Current statistical data (count, min, max values)
Basic authentication support for secured Prometheus instances

Configuration

The server requires the following environment variable:

PROMETHEUS_URL: The base URL of your Prometheus instance

Optional authentication configuration:

PROMETHEUS_USERNAME: Username for basic auth (if required)
PROMETHEUS_PASSWORD: Password for basic auth (if required)

Development

Install dependencies:


npm install

Build the server:


npm run build

For development with auto-rebuild:


npm run watch

Installation

To use with Claude Desktop, add the server config:

On MacOS: ~/Library/Application Support/Claude/claude_desktop_config.json On Windows: %APPDATA%/Claude/claude_desktop_config.json


{
  "mcpServers": {
    "mcp-server-prometheus": {
      "command": "/path/to/mcp-server-prometheus/build/index.js",
      "env": {
        "PROMETHEUS_URL": "http://your-prometheus-instance:9090"
      }
    }
  }
}

Debugging

Since MCP servers communicate over stdio, debugging can be challenging. We recommend using the MCP Inspector :


npm run inspector

The Inspector will provide a URL to access debugging tools in your browser.

API Structure

The server exposes Prometheus metrics through the following URI structure:

Base URI: http://your-prometheus-instance:9090
Metric URIs: http://your-prometheus-instance:9090/metrics/{metric_name}

Each metric resource returns JSON data containing:

Metric name
Metadata (help text, type)
Current statistics (count, min, max)