Airflow Extension for Visual Studio Code
✅ Now works with Airflow 3.0
🤖 Chat with Airflow
A Visual Studio Code extension to interact with Apache Airflow (v2.x & v3.x) directly from your editor.
Trigger DAGs, pause/unpause, explore DAG runs, view logs, browse code, and more—no browser required.
🔑 Features
The extension provides 24 language model tools that integrate with VS Code's AI chat. Use @airflow in the chat to access these tools.
Trigger DAG Run
- Purpose: Execute a DAG with optional configuration
- Sample Prompts:
@airflow trigger data_pipeline_dag@airflow trigger etl_job_dag with config {"max_workers": 10}@airflow run my_dag on 2025-12-03
Pause DAG
- Purpose: Stop new DAG runs from being scheduled
- Sample Prompts:
@airflow pause data_pipeline_dag@airflow disable staging_etl_dag@airflow pause all failing dags
Unpause DAG
- Purpose: Enable scheduling for a paused DAG
- Sample Prompts:
@airflow unpause data_pipeline_dag@airflow activate my_dag@airflow resume production_job
Stop DAG Run
- Purpose: Terminate a currently running DAG execution
- Sample Prompts:
@airflow stop data_pipeline_dag run@airflow cancel current run of my_dag@airflow kill the running data_ingestion_dag
Get DAG Runs
- Purpose: View execution history for a specific DAG with optional date filter
- Sample Prompts:
@airflow get dag runs for api_data_ingestion_dag@airflow show runs of data_pipeline_dag on 2025-12-03@airflow list runs for etl_job_dag from last week
Get DAG History
- Purpose: View DAG run history on a specific date with status, duration, and notes
- Sample Prompts:
@airflow get dag history of api_data_ingestion_dag@airflow show history for data_pipeline_dag on 2025-12-02@airflow view past runs of my_dag today
Get Failed Runs
- Purpose: Identify and analyze failed DAG runs across the system
- Sample Prompts:
@airflow get failed runs@airflow show failed dags from last 24 hours@airflow which dags failed in the last 48 hours
List Active DAGs
- Purpose: View all currently enabled DAGs
- Sample Prompts:
@airflow list active dags@airflow show all running dags@airflow get enabled dags
List Paused DAGs
- Purpose: View all currently disabled DAGs
- Sample Prompts:
@airflow list paused dags@airflow show disabled dags@airflow get all paused dags
Get Running DAGs
- Purpose: List all DAGs that currently have running or queued tasks
- Sample Prompts:
@airflow get running dags@airflow get active runs@airflow what is running now?
Analyse DAG Latest Run
- Purpose: Comprehensive diagnostic report of the latest DAG execution including tasks, logs, and source code
- Sample Prompts:
@airflow analyze api_data_ingestion_dag@airflow get full diagnostics for data_pipeline_dag@airflow show latest execution details and logs for etl_job_dag@airflow what went wrong with my_dag last run
Get DAG Run Detail
- Purpose: Deep dive analysis of a specific DAG run ID
- Sample Prompts:
@airflow analyze run <run_id> of <dag_id>@airflow get run details for dag_id=my_dag run_id=manual__2023...
Get DAG Source Code
- Purpose: Retrieve and view the Python source code of a DAG
- Sample Prompts:
@airflow review source code for data_pipeline_dag@airflow show source code for data_pipeline_dag@airflow get code for my_etl_dag@airflow let me see the implementation of api_ingestion_dag
AI Chat Analysis
- Purpose: Get AI-powered insights and recommendations using all available tools
- Sample Prompts:
@airflow why is data_pipeline_dag failing? Analyze and show me the logs@airflow give me a summary of failed runs and recommend fixes@airflow check if my dags are healthy and pause any that keep failing
Open Views
- Purpose: Quickly navigate to any part of the extension using natural language
- Sample Prompts:
@airflow open log view for my_dag@airflow show me variables@airflow go to connections@airflow open provider list@airflow show server health
� MCP Server (Model Context Protocol)
The extension includes a built-in MCP bridge server that exposes all Airflow tools to any external AI agent or client that supports the Model Context Protocol (e.g., Cursor, Windsurf, Antigravity, Claude Desktop, Continue, and other MCP-compatible tools).
How It Works
- The MCP bridge listens on
127.0.0.1:37115 (configurable).
- An external MCP client connects and discovers all available Airflow tools.
- The client can call any tool (trigger DAG, get logs, analyze runs, etc.) just as
@airflow does in VS Code Chat.
Getting Started with MCP
- Open the Airflow sidebar and expand the MCP section.
- Click Start to launch the bridge server.
- Click Manage to open the MCP Manager and copy the generated config snippet.
- Paste the config snippet into your MCP client's configuration file.
MCP Manager
Open the MCP Manager via the sidebar (MCP → Manage) to:
- View the current server host and port.
- Copy the ready-to-paste JSON config for your MCP client.
- Start or stop the bridge server.
Example Config (Cursor / Windsurf / Antigravity / Claude Desktop)
{
"mcpServers": {
"airflow": {
"command": "node",
"args": ["/path/to/extension/out/mcp/cli.js"],
"env": {
"AIRFLOW_MCP_HOST": "127.0.0.1",
"AIRFLOW_MCP_PORT": "37115"
}
}
}
}
Tip: The MCP Manager generates the exact args path for your machine — just copy and paste it.
| Action |
Description |
| Status |
Show current server state (running / stopped) and active session count |
| Start |
Start the MCP bridge server |
| Stop |
Stop all active MCP sessions and the bridge server |
| Manage |
Open the MCP Manager view with config snippet and settings |
�📷 Screenshots
| Daily Dag Runs |
Dag Run History |
 |
 |
⚙️ Configuration
After installing the extension, you need to configure the extension to connect to your Airflow Server.
- Click 🔌 Connect to Airflow Server at the top of the extension sidebar.
- Enter your Airflow Server API Url
- Exp Airflow 2: http://localhost:8080/api/v1
- Exp Airflow 3: http://localhost:8080/api/v2
- Enter your Airflow username
- Enter your Airflow password
You can also add multiple Airflow Servers to connect to.
- Use [+] button to add a new Airflow Server
- Use [-] button to remove an Airflow Server
- Use [🔌] to switch between servers
ℹ️ To be able to connect an Airflow Server, you should enable Airflow Rest Api.
You can take a look the link below on how to do it.
https://airflow.apache.org/docs/apache-airflow/stable/security/api.html
🪲 Bug & New Feature Report
If you have an issue or new feature request, please click link below to add a new issue.
https://github.com/necatiarslan/airflow-vscode-extension/issues/new
💻 Local Airflow via Astronomer
Use Astro provided by the team who build Airflow
https://docs.astronomer.io/astro/cli/overview
To spin up a local Airflow instance for testing:
brew install astro #install cli tool from homebrew
astro dev init #init the local env
astro dev start #start airflow as a docker container
#Then you can connect your local airflow using the extension.
#airflow 2 url http://localhost:8080/api/v1
#airflow 3 url http://localhost:8080/api/v2
#user/pass admin/admin
📊 Telemetry & Privacy
This extension collects anonymous usage telemetry to help improve reliability and features.
Telemetry control
Telemetry is collected only if VS Code telemetry is enabled.
You can disable telemetry via: Settings → Telemetry → Telemetry Level
📝 Roadmap
Coming Soon
- Task View
- Instance Details
- Rendered Template
- Test Connection language model tool
- Add trigger config files
If you find this extension useful, please consider:
📬 Stay in Touch
Enjoy! 🚀
