Airflow Token

Apache Airflow MCP server with Bearer token authentication support for Astronomer and standalone Airflow

python

GitHub Stars

python

Language

6 months ago

First Indexed

2 months ago

Catalog Refreshed

Documentation & install

Readme and setup notes from the catalogue, plus a client-ready config you can copy for your MCP host.

Installation

Add the following to your MCP client configuration file.

Configuration

View docs

{
  "mcpServers": {
    "nikhil-ganage-mcp-server-airflow-token": {
      "command": "uv",
      "args": [
        "--directory",
        "path-to-repo/mcp-server-airflow-token",
        "run",
        "mcp-server-airflow-token"
      ],
      "env": {
        "AIRFLOW_HOST": "<airflow-host>",
        "AIRFLOW_TOKEN": "<airflow-token>",
        "AIRFLOW_PASSWORD": "your-password",
        "AIRFLOW_USERNAME": "your-username",
        "AIRFLOW_API_VERSION": "v1"
      }
    }
  }
}

You run an MCP server that wraps Apache Airflow’s REST API and adds token-based authentication to securely connect from clients. This server supports Bearer token authentication (recommended), is compatible with Astronomer Cloud deployments, and remains backward compatible with username/password authentication. It exposes a wide set of Airflow endpoints through a standardized MCP interface, enabling you to manage DAGs, DAG runs, variables, connections, and more from a consistent client experience.

How to use

You interact with the MCP server from your MCP client. The server communicates with Airflow, authenticating requests using either a Bearer token or username/password, depending on what you supply. To run in token mode, provide your Airflow API token. To run in basic mode, provide Airflow username and password. You can choose to run in read-only mode to restrict changes while still accessing read operations like listing DAGs, DAG runs, variables, and configurations.

How to install

Prerequisites you need on your system before starting:

Install the MCP server package using Python’s package manager or run it with the included runtime tool.

Additional sections

Configuration and usage patterns are shown via concrete examples you can adapt to your environment. The server exposes a comprehensive set of Airflow endpoints such as DAGs, DAG runs, tasks, variables, connections, pools, XComs, datasets, and monitoring health. You can selectively enable API groups when starting the server to tailor the surface you expose to clients.

Security and access

Bearer token authentication is the primary method for modern Airflow deployments. If AIRFLOW_TOKEN is provided, it will be used for authentication; otherwise, the server will fall back to basic authentication using AIRFLOW_USERNAME and AIRFLOW_PASSWORD.

Available tools

dag

Operations to manage DAGs: list, get details, pause/unpause, update, delete, patch multiple, and other DAG-related actions.

dagrun

Operations to manage DAG runs: list, create, get details, update, delete, and batch retrieval.

taskinstance

Operations to manage task instances: get, list, update, clear, and set state.

variable

Operations to manage Airflow variables: list, create, get, update, delete.

connection

Operations to manage connections: list, create, get, update, delete, test.

pool

Operations to manage pools: list, create, get, update, delete.

xcom

Operations to access XCom entries: list and get specific entries.

datasets

Operations to manage datasets and dataset events: list, get, create, delete.

eventlog

Operations to access event logs: list and get details.

importerror

Operations to access import errors: list and get details.

health

Health and version endpoints to monitor the MCP server and Airflow integration.

config

Access configuration details via the config endpoint.

plugin

List and inspect Airflow plugins exposed through the MCP surface.

provider

List and inspect Airflow providers available via MCP.

monitoring

Access monitoring endpoints including health status and statistics.

dagstat

Retrieve statistics for DAGs, such as execution metrics.

Built by

VeilStrat

AI signals for GTM teams