Airflow MCP Server

Example Use Cases

• 1

  List and summarize DAGs, their schedules, and recent run statuses from an Airflow cluster.

• 2

  Diagnose and troubleshoot failed DAG runs and retrieve task-level logs and metadata.

• 3

  Programmatically trigger DAG runs, monitor execution results, and integrate Airflow visibility into conversational workflows or LangChain agents.

Description

This project implements a Model Context Protocol (MCP) server for Apache Airflow, exposing Airflow metadata and control actions (e.g., listing DAGs, checking run status, triggering runs) to LLMs and agents. It includes a prebuilt Docker image and an optional bundled local Airflow for demo purposes, plus integration examples for Claude Desktop and LangChain. The server simplifies operational workflows by enabling natural-language troubleshooting, monitoring, and automated intervention across Airflow pipelines.