Apache Airflow MCP MCP Server

About

## Airflow MCP Server: Workflow Orchestration The **Airflow MCP Server** integrates Apache Airflow's workflow platform into Google Antigravity. This enables building, scheduling, and monitoring data pipelines with DAGs (Directed Acyclic Graphs). ### Why Airflow MCP? Airflow orchestrates complex workflows: - **DAGs**: Define workflows as code - **Scheduling**: Cron-like scheduling - **Monitoring**: Rich UI for pipeline tracking - **Extensible**: 1000+ operators and hooks - **Scalable**: Distributed execution ### Key Features #### 1. Trigger DAGs ```python import requests headers = {"Authorization": f"Bearer {api_token}"} # Trigger DAG run response = requests.post( "https://airflow.example.com/api/v1/dags/etl_pipeline/dagRuns", headers=headers, json={ "conf": {"date": "2024-03-01"}, "logical_date": "2024-03-01T00:00:00Z" } ) run_id = response.json()["dag_run_id"] ``` #### 2. Monitor Runs ```python # Get DAG run status response = requests.get( f"https://airflow.example.com/api/v1/dags/etl_pipeline/dagRuns/{run_id}", headers=headers ) run = response.json() print(f"State: {run['state']}") # Get task instances tasks = requests.get( f"https://airflow.example.com/api/v1/dags/etl_pipeline/dagRuns/{run_id}/taskInstances", headers=headers ) ``` #### 3. Manage DAGs ```python # Pause/unpause DAG requests.patch( "https://airflow.example.com/api/v1/dags/etl_pipeline", headers=headers, json={"is_paused": False} ) # List DAGs dags = requests.get( "https://airflow.example.com/api/v1/dags", headers=headers ) ``` ### Configuration ```json { "mcpServers": { "airflow": { "command": "npx", "args": ["-y", "@anthropic/mcp-airflow"], "env": { "AIRFLOW_HOST": "https://airflow.example.com", "AIRFLOW_USERNAME": "admin", "AIRFLOW_PASSWORD": "password" } } } } ``` ### Use Cases **ETL Pipelines**: Extract, transform, load workflows. **ML Pipelines**: Model training and deployment. **Data Quality**: Scheduled data validation. The Airflow MCP Server brings workflow orchestration to Antigravity.

Installation

{
  "mcpServers": {
    "airflow": {}
  }
}

How to Use