Airflow MCP Server for Model Context Protocol
Integrate Apache Airflow with MCP clients using this MCP server for standardized Model Context Protocol interactions.
npx -y @yangkyeongmo/mcp-server-apache-airflowOverview
The Airflow MCP Server implements a Model Context Protocol (MCP) server that fronts an Apache Airflow deployment. It exposes a consistent, MCP-friendly interface that maps common MCP actions to Airflow’s REST API endpoints, so MCP clients and orchestration-aware tooling can interact with Airflow in a standardized way.
This server uses the official Airflow REST client under the hood and forwards MCP requests to the corresponding Airflow endpoints (DAG management, DAG runs, tasks, variables, etc.). That makes it useful when you want a single protocol-compliant surface for model-context tooling, observability agents, or other components that speak MCP to control and inspect Airflow.
Features
- Standardized MCP wrapper around the Airflow REST API
- Full coverage of common DAG, DAG run, task, and variable operations
- Uses the official Apache Airflow client for compatibility
- Simple configuration via environment variables
- Can be deployed alongside existing Airflow instances or in front of an Airflow REST endpoint
Notable implemented mappings (MCP actions → Airflow REST API):
| Area | Example API path |
|---|---|
| DAGs | /api/v1/dags |
| DAG details | /api/v1/dags/{dag_id} |
| DAG runs | /api/v1/dags/{dag_id}/dagRuns |
| Task ops | /api/v1/dags/{dag_id}/tasks/{task_id} |
| Task logs | /api/v1/dags/{dag_id}/dagRuns/{run}/logs |