airflow-mcp-server logo

airflow-mcp-server

by abhishekbhakat

An MCP (Model Context Protocol) server for controlling Airflow via Airflow APIs. It allows interaction with Airflow instances using the Model Context Protocol.

View on GitHub

Last updated: N/A

What is airflow-mcp-server?

This is a Model Context Protocol (MCP) server designed to control and interact with Apache Airflow instances through their APIs. It acts as a bridge, enabling communication and control of Airflow workflows using the MCP standard.

How to use airflow-mcp-server?

To use the server, configure it with the Airflow API base URL and authentication credentials (either a base64 encoded username:password or a session cookie) via environment variables. You can then start the server in either safe (read-only) or unsafe (full access) mode. The provided JSON configuration shows how to integrate it with Claude Desktop.

Key features of airflow-mcp-server

  • Controls Airflow via Airflow APIs

  • Supports Safe and Unsafe operation modes

  • Supports Basic Auth and Cookie authentication

  • Parses OpenAPI Spec

  • Configurable page limit

Use cases of airflow-mcp-server

  • Automating Airflow workflow management

  • Integrating Airflow with MCP-compatible tools

  • Remotely controlling Airflow deployments

  • Building custom Airflow interfaces

  • Providing a standardized interface for Airflow interaction

FAQ from airflow-mcp-server

What is the Model Context Protocol (MCP)?

MCP is a standard protocol for interacting with models and related services.

How do I authenticate with the Airflow API?

You can authenticate using either a base64 encoded username:password via the AUTH_TOKEN environment variable or a session cookie via the COOKIE environment variable.

What is Safe Mode?

Safe Mode only allows read-only operations (GET requests), preventing any modifications to your Airflow instance.

What environment variables are required?

The required environment variables are AIRFLOW_BASE_URL, and either AUTH_TOKEN or COOKIE.

How can I change the page limit?

You can change the page limit using the maximum_page_limit option in the [api] section of the airflow.cfg file.