us-all/airflow-mcp-server

What it does

Airflow MCP Server is a Model Context Protocol (MCP) server that bridges Apache Airflow with AI assistants like Claude. It exposes Airflow's REST API v2 through MCP tools, allowing AI models to list DAGs, view runs and task instances, tail logs, trigger DAGs, and clear task instances. Includes DAG health rollup aggregation for monitoring workflow health status.

How to set up

Install the MCP server and configure it to connect to your Airflow instance. Authentication uses JWT with SimpleAuthManager. For Airflow 3.x, use the latest version with api/v2 support. For Airflow 2.x, pin to version 0.1.x. Add the server to your Claude or AI assistant MCP configuration to enable tool access.

Key features

• 7 integrated tools: DAG list, runs, task instances, log tails, trigger DAGs, clear tasks, and health rollup
• Read-only by default with optional write-gated access for trigger and clear operations
• JWT authentication support with SimpleAuthManager
• Compatible with Airflow 3.x (api/v2) and 2.x versions
• DAG health aggregation for workflow monitoring
• Direct integration with Claude and MCP-compatible AI assistants