Mcp Airflow Postgres

What is Mcp Airflow Postgres?

Mcp Airflow Postgres is a Model Context Protocol (MCP) server that connects to an Airflow PostgreSQL database, allowing users to query failed DAG runs and execute custom SQL. It runs locally via Python and is designed for developers and data engineers who want to integrate Airflow database interactions into MCP-compatible tools.

How to use Mcp Airflow Postgres?

Install Python 3.8+ and Poetry, clone the repository, run poetry install, create a .env file with a DATABASE_URL pointing to your Airflow PostgreSQL instance, then start the server with poetry run python src/server.py. The server can also be used with Smithery by providing the DATABASE_URL configuration as defined in smithery.yaml.

Key features of Mcp Airflow Postgres

• Connects to Airflow PostgreSQL databases via MCP
• Query failed DAG runs within a specified time period
• Execute arbitrary SQL queries directly
• Simple setup using Poetry and environment variables
• Compatible with the Smithery platform

Use cases of Mcp Airflow Postgres

• Monitor and inspect failed Airflow DAG runs programmatically
• Run ad‑hoc SQL queries against the Airflow metadata database
• Integrate Airflow database access into AI assistants or automation tools
• Automate Airflow health checks within MCP workflows

FAQ from Mcp Airflow Postgres

What does Mcp Airflow Postgres do?

It is an MCP server that lets you query failed DAG runs and execute direct SQL on an Airflow PostgreSQL database.

Can I use it with Smithery?

Yes, the server is configured for Smithery via smithery.yaml; you must supply the DATABASE_URL when starting.

What are the available tools?

Two tools: failed_runs – queries failed DAG runs within a given time period; query – runs arbitrary SQL against the database.

What database does it support?

PostgreSQL, as shown in the example connection string postgresql://airflow:airflow123@localhost:5432/airflow.

How do I install and run it?

Requires Python 3.8+ and Poetry. Clone the repo, run poetry install, set a DATABASE_URL in .env, then run poetry run python src/server.py.