Prometheus

Provides access to Prometheus data through MCP endpoints for retrieval, analysis, and complex queries.

python

0

GitHub Stars

python

Language

7 months ago

First Indexed

3 months ago

Catalog Refreshed

Documentation & install

Readme and setup notes from the catalogue, plus a client-ready config you can copy for your MCP host.

Installation

Add the following to your MCP client configuration file.

Configuration

View docs

{
  "mcpServers": {
    "mcp-mirror-caesaryangs_prometheus_mcp_server": {
      "command": "uv",
      "args": [
        "--directory",
        "/path/to/prometheus_mcp_server",
        "run",
        "server.py"
      ],
      "env": {
        "PROMETHEUS_HOST": "http://localhost:9090"
      }
    }
  }
}

You can run an MCP server that fetches and analyzes data from Prometheus, enabling large language models to retrieve metrics, perform analyses, search usage patterns, and execute advanced queries through clearly defined tool endpoints.

How to use

You interact with the Prometheus MCP server by starting it in stdio mode and then issuing requests through your MCP client. The server exposes core capabilities such as retrieving metric information, fetching metric data, analyzing data over time ranges, and performing complex queries. You can connect your LLM workflow to fetch metrics, analyze results, and explore metric usage without directly querying Prometheus yourself.

How to install

Prerequisites you need before starting: a Python-enabled environment and, if you plan to run the client integration, a compatible MCP client setup.

cd ./src/prometheus_mcp_server
python3 -m venv .venv

# linux/macos:
source .venv/bin/activate

# windows:
.venv\Scripts\activate

Additional setup steps

Install Python packaging tools if they are not present and install required dependencies for the MCP server.

# ensure pip is available in your venv
wget https://bootstrap.pypa.io/get-pip.py
python3 get-pip.py

# install dependencies
pip install -r requirements.txt

Starting the MCP server

You have two options to start the Prometheus MCP server in stdio mode. Use your preferred method and ensure the environment is set up as described.

# Option 1: Claude/UV method (stdio) using the project path you choose
uv --directory /path/to/prometheus_mcp_server run server.py

# Option 2: Direct Python start (stdio) with the inline server file
python3 server.py

Configuration basics

The MCP server reads its host configuration from environment variables when started in stdio mode. The following variable is used by the CLI to connect to the Prometheus instance.

{
  "PROMETHEUS_HOST": "http://localhost:9090"
}

Available tools

data_retrieval

Fetch metric data by name and time ranges from Prometheus, enabling you to retrieve specific metrics or ranges for analysis.

metric_analysis

Analyze retrieved metric data statistically, summarize trends, and compute basic statistics over chosen time windows.

usage_search

Search metric usage patterns and connections across Prometheus metrics to identify relationships and hotspots.

complex_querying

Execute advanced PromQL queries to explore in-depth data and produce detailed results for deeper insights.