Overview
What is Prometheus MCP Server?
A tool that allows access to Prometheus data through a Model Context Protocol server. It is intended for users who need to query Prometheus metrics from MCP‑compatible clients like Claude.
How to use Prometheus MCP Server?
Install automatically via Smithery, manually with pipx install, or run without installation using pipx run. Configure with command‑line arguments: --url (required), optional --username/--password, --token, --org-id, --timeout, and --limit.
Key features of Prometheus MCP Server
- Supports basic authentication (username/password) and token authentication
- Configurable evaluation timeout and maximum series limit
- Multi‑tenant support via organization ID (
--org-id) - Can be run without permanent installation using
pipx run - Provides direct access to Prometheus data through the MCP protocol
Use cases of Prometheus MCP Server
- Query Prometheus metrics from within AI assistants that support MCP
- Perform ad‑hoc analysis of monitoring data without a permanent install
- Integrate Prometheus monitoring data into automated workflows
FAQ from Prometheus MCP Server
What authentication methods are supported?
Basic authentication with --username/--password or token authentication with --token.
How do I specify the Prometheus server URL?
Use the --url command‑line argument with the full URL (e.g., http://your-prometheus-server:9090).
What are the default values for --timeout and --limit?
The default timeout is 30 seconds; the default limit is 1000 returned series.
Can I use the server without installing it?
Yes, by running pipx run --spec git+https://github.com/kakao-yanoo-kim/prometheus-mcp-server-py.git prometheus-mcp --url ....
Does the server support multi‑tenant Prometheus?
Yes, by providing the --org-id argument for the organization ID.
