Log inSubmit server
Airflow MCP Server logo

Airflow MCP Server

by us-all

Connects MCP clients to Airflow 3.x for DAG inspection, health analysis, log tails, and gated run management.

InstallRepositoryWebsite
Full README Compare alternatives Embed widget Claim this listing

Install Airflow MCP Server

Pick your client, then add it in one click or paste the config.

STDIONode.js
~/Library/Application Support/Claude/claude_desktop_config.json
claude_desktop_config.json
{
  "mcpServers": {
    "us-all-airflow-mcp-server": {
      "command": "npx",
      "args": [
        "-y",
        "@us-all/airflow-mcp"
      ],
      "env": {
        "AIRFLOW_API_URL": "YOUR_VALUE_HERE",
        "AIRFLOW_USERNAME": "YOUR_VALUE_HERE",
        "AIRFLOW_PASSWORD": "YOUR_VALUE_HERE",
        "AIRFLOW_ALLOW_WRITE": "YOUR_VALUE_HERE",
        "AIRFLOW_TOOLS": "YOUR_VALUE_HERE",
        "AIRFLOW_DISABLE": "YOUR_VALUE_HERE"
      }
    }
  }
}

Settings → Developer → Edit Config, merge this into mcpServers, then fully quit and reopen Claude Desktop.

Values are written into the snippet above so you can paste it as is. They never leave your browser.

We couldn’t confirm it starts correctly

We ran the install command but it didn't respond within our test window — this can mean a slow first-time install rather than a real problem.

npx -y @us-all/airflow-mcp

No response to initialize.

This is an experimental automated check and can have false negatives — missing environment variables, a slow cold install, etc. It doesn’t necessarily mean something’s wrong. Last checked 1mo ago.

Overview

The us-all/airflow-mcp-server MCP server exposes Airflow 3.x’s REST API through MCP tools for browsing DAGs, inspecting runs and task instances, reading log tails, and summarizing DAG health. It authenticates with Airflow’s SimpleAuthManager by minting and refreshing JWT tokens from a username and password. Read operations are enabled by default, while triggering DAG runs and clearing tasks require an explicit write setting. Reach for it when an agent needs operational visibility into Airflow workflows or controlled run remediation from an MCP client.

Use cases

  • Inspect recent Airflow DAG failures
  • Review task logs and failing instances
  • Summarize DAG health across runs
  • Trigger or re-run tasks with write access enabled

Key features

  • Airflow 3.x REST API support
  • DAG, run, task, and log inspection
  • Aggregated DAG health rollups
  • JWT authentication with token refresh
  • Write-gated triggering and task clearing
  • MCP stdio and optional Streamable HTTP transport

Tools7

What Airflow MCP Server lets your agent do. Expand a tool for its parameters.

Self-reported~120 tokens of context

Security & health

What it can reach and how reliably it runs — from our own checks, not the maintainer’s claims.

ActiveLast checked Today

Recent health check succeeded.

17/18 checks healthy over the last 78d

Supply-chain signal

No high-severity advisories surfaced by our automated scan.

Critical 0High 0Medium 0Low 0

Scanned 8d ago via OSV.dev · @us-all/airflow-mcp (npm)

Adoption & maintenance

Factual signals from GitHub, npm, and our automated checks — not a rating.

GitHub stars2Stargazers on the source repository.
npm downloads214Package downloads in the last 30 days.
Last commit3mo agoMost recent push to the default branch.
Install checkInconclusiveDidn't respond in our test window — often a slow first install.
Tools exposed7Callable tools this server registers over MCP.
Directory activity7 viewsConfig copies, upvotes, and views on AllMCPs.

How Airflow MCP Server works

What us-all/airflow-mcp-server does

The us-all/airflow-mcp-server MCP server connects an MCP client to the Airflow 3.x REST API at /api/v2. It provides tools for finding active DAGs, reviewing recent runs, inspecting task instances, and retrieving the end of a task log. A health rollup combines operational signals into a DAG-level view with success rate, average duration, the most recent failed run, and failing tasks.

The server also includes two workflow prompts. dag-failure-triage gathers run and task information, retrieves relevant logs, and proposes remediation for a failed DAG. dag-schedule-audit checks DAGs for low success rates and stuck runs. The search-tools meta tool supports natural-language discovery of available tools.

How it works

At startup, the server receives an Airflow host base URL and adds /api/v2 internally. If the supplied URL ends in /api/v1 or /api/v2, that suffix is removed first. Authentication uses Airflow SimpleAuthManager: the server sends the configured credentials to /auth/token, caches the resulting JWT for the session, and refreshes it one minute before expiration.

The us-all/airflow-mcp-server MCP server uses MCP stdio by default, so it can be configured in local clients such as Claude Desktop and Cursor. Setting MCP_TRANSPORT=http switches to Streamable HTTP transport, which provides Bearer authentication and a /health endpoint.

Setup and configuration

Install the npm package as a development dependency with pnpm, or run the published package directly with npx. The server requires AIRFLOW_API_URL, AIRFLOW_USERNAME, and AIRFLOW_PASSWORD. The URL should identify the Airflow host rather than include an API path.

Set AIRFLOW_ALLOW_WRITE=true only when the client should be able to use airflow-trigger-dag and airflow-clear-task. Without that setting, the server remains read-only for those operations. AIRFLOW_TOOLS and AIRFLOW_DISABLE are available for category-level tool toggles.

Tools and capabilities

  • List active DAGs with tag and search filters.
  • Retrieve recent runs for a selected DAG, optionally filtered by state and ordered newest first.
  • Inspect task instances for a specific DAG run.
  • Read the last N kilobytes of a task instance log.
  • Trigger a DAG run or clear selected task instances when writes are enabled.
  • Produce an aggregated health report for a DAG.

Limitations and notes

The us-all/airflow-mcp-server MCP server targets Airflow 3.0–3.2 and the /api/v2 API; the README reports verification against Airflow 3.2.1. Version 0.2 and later do not support Airflow 2.x deployments because the required basic-auth and /api/v1 behavior was removed. Airflow 2.x users should use @us-all/airflow-mcp@0.1.0 instead.

This package does not provide the deeper dbt artifact, lineage, source-freshness, or data-quality capabilities described for the companion @us-all/dbt-mcp server. The HTTP transport is opt-in; stdio is the default transport.

Read the full README View source on GitHub

Reviews

No reviews yet — be the first to share how it worked for you.

Frequently Asked Questions about Airflow MCP Server

Run `npx -y @us-all/airflow-mcp` or add `@us-all/airflow-mcp` as a development dependency with pnpm.

Yes.

Versions 0.2 and later target Airflow 3.0–3.2 through `/api/v2`. For Airflow 2.x, use `@us-all/airflow-mcp@0.1.0`.

Set `AIRFLOW_ALLOW_WRITE=true`. This enables the trigger-DAG and clear-task operations; read operations are enabled by default.

Related MCP servers

All in Data Platforms All alternatives
Add a badge to your READMEShow it’s listed — live SVG, Markdown or HTML.

AllMCPs Directory Badge

Full Badge Customizer

Showcase your server listing on GitHub or your project documentation. Embed this dynamic SVG badge to highlight official listing status and live engagement.

Badge Style:
Live Dynamic SVG PreviewAirflow MCP Server AllMCPs Directory Badge
Markdown (GitHub README)
[![AllMCPs](https://allmcps.com/api/badge/us-all-airflow-mcp-server?style=directory)](https://allmcps.com/mcp/us-all-airflow-mcp-server)
HTML Embed
<a href="https://allmcps.com/mcp/us-all-airflow-mcp-server"><img src="https://allmcps.com/api/badge/us-all-airflow-mcp-server?style=directory" alt="Airflow MCP Server on AllMCPs" /></a>