prometheus-mcp-server
by tjhop
Overview
This MCP server allows LLMs to interact with Prometheus instances via its API for monitoring, querying, and analysis tasks.
Installation
docker run --rm -p 8080:8080 ghcr.io/tjhop/prometheus-mcp-server:latest --prometheus.url "http://localhost:9090" --mcp.transport "http" --web.listen-address ":8080"Environment Variables
- PROMETHEUS_MCP_SERVER_PROMETHEUS_URL
Security Notes
The server's code quality appears high, utilizing standard Go libraries and Prometheus client-golang. A critical security consideration is the `--dangerous.enable-tsdb-admin-tools` flag, which, if enabled, exposes destructive operations (like deleting time-series data) to the LLM. While this is explicitly opt-in with warnings, it represents a significant potential risk. The `apiClientLoaderMiddleware` handles and proxies `Authorization` headers to the Prometheus backend, making securing the MCP server's own endpoints paramount. Network calls to the Prometheus instance are made based on a configured URL, reducing the direct risk of LLM-induced arbitrary network requests from within tool handlers, but the target Prometheus instance must be trusted.
Similar Servers
mcp-grafana
Provides a Model Context Protocol (MCP) server to access Grafana instances and its ecosystem for observability and incident management tasks.
bifrost
High-performance AI gateway unifying multiple LLM providers, offering real-time monitoring, configuration management, and extensibility through plugins.
prometheus-mcp-server
Enables AI assistants to query and analyze Prometheus metrics through a Model Context Protocol (MCP) server.
logfire-mcp
Enables LLMs to retrieve and analyze application telemetry data (OpenTelemetry traces and metrics) from Pydantic Logfire using SQL queries.