This repository provides a Model Context Protocol (MCP) server implementation for Apache Airflow, enabling standardized interaction with Airflow through the MCP. It wraps Airflow's REST API, allowing MCP clients to
This project implements a Model Context Protocol (MCP) server for Apache Airflow, facilitating standardized interaction with Airflow through MCP clients. It leverages the official Apache Airflow client library for compatibility.
Feature Implementation Status:
The server provides comprehensive coverage of Airflow's REST API, including DAG Management (listing, details, pausing, unpausing, updating, deleting, source retrieval, reparsing), DAG Runs (listing, creation, details, updating, deleting, batch retrieval, clearing, setting notes, upstream dataset events), Tasks (listing, details, instances management), Variables, Connections, Pools, XComs, Datasets, Monitoring, DAG Stats, Config, Plugins, Providers, Event Logs, and System information.
Setup:
Requires apache-airflow-client
. Environment variables AIRFLOW_HOST
, AIRFLOW_USERNAME
, and AIRFLOW_PASSWORD
must be set. Configuration examples for Claude Desktop using uvx
and uv
are provided. API groups can be selected using the --apis
flag. Manual execution via make run
or make run-sse
is also supported. Installation via Smithery is available.
Contributing:
Contributions are welcome.
License:
MIT License
yangkyeongmo/mcp-server-apache-airflow
February 13, 2025
March 27, 2025
Python