Grafana-Loki MCP Server

What is Grafana-Loki MCP Server?

It is a FastMCP server that allows querying Loki logs from a Grafana instance. It is intended for developers and operations teams who need to interact with Loki log data programmatically through the Model Context Protocol.

How to use Grafana-Loki MCP Server?

Install via pip install grafana-loki-mcp or run with uvx grafana-loki-mcp. Configure it by setting the GRAFANA_URL and GRAFANA_API_KEY environment variables, or pass them as command-line arguments (-u and -k). The server supports both stdio and SSE transport protocols.

Key features of Grafana-Loki MCP Server

• Query Loki logs through the Grafana API
• Retrieve Loki label names and label values
• Format query results as text, JSON, or markdown
• Supports stdio and SSE transport protocols
• Accepts Grafana-style relative time ranges (e.g., now-1h)
• Limits log line length per entry (configurable max per line)

Use cases of Grafana-Loki MCP Server

• Integrate Loki log queries into MCP‑compatible AI assistants and tools
• Automate log analysis pipelines by fetching and formatting logs
• Programmatically retrieve label metadata for log exploration
• Build custom log dashboards or alerts using external scripts

FAQ from Grafana-Loki MCP Server

What are the runtime requirements?

Python 3.10 or higher, plus the FastMCP and Requests libraries.

How do I configure the server with Grafana?

Provide your Grafana instance URL and API key either via environment variables (GRAFANA_URL, GRAFANA_API_KEY) or with command-line arguments -u and -k.

What transport protocols does the server support?

It supports both stdio (default) and sse (set with --transport sse).

How can I query logs and format the results?

Use the query_loki tool with parameters such as query, start, end, limit, direction, and max_per_line. Pass the results to format_loki_results to output them in text, JSON, or markdown.