Run, monitor, and debug DAGs straight from Microsoft Teams using natural language.
661
This guide explains how to run the Airflow Copilot using Docker.
The Airflow Copilot is a GenAI-powered assistant that interacts with Apache Airflow environments. You can run it easily using Docker, either locally or in the cloud.
For in-depth configuration and deployment instructions, refer to:
docker pull thedatacarpenter/airflow-copilot:<version>
docker pull thedatacarpenter/airflow-copilot:latest
docker run --rm -p 3978:3978 thedatacarpenter/airflow-copilot:latest
The container requires several environment variables, categorized below:
LLM_MODEL_PROVIDER_NAME=OpenAI
LLM_MODEL_NAME=gpt-4o
OPENAI_API_KEY=your-openai-api-key
Other providers:
GOOGLE_GENAI_API_KEY=ai-...
ANTHROPIC_API_KEY=ai-...
GROQ_API_KEY=ai-...
MICROSOFT_APP_ID=your-bot-id
MICROSOFT_APP_PASSWORD=bot-secret
AZURE_CLIENT_ID=your-client-id
AZURE_CLIENT_SECRET=your-spn-secret
AZURE_TENANT_ID=your-tenant-id
RESOURCE_GROUP=your-resource-group
AIRFLOW_BASE_URL=http://localhost:8080/
AIRFLOW_AUTH_STRATEGY=per_user
DB_URI=postgresql://<user>:<password>@<host>:<port>/<database>
FERNET_SECRET_KEY=your-fernet-secret-key
Generate with:
python -c "from cryptography.fernet import Fernet; print(Fernet.generate_key().decode())"
MIN_MSG_TO_RETAIN=10
MIN_MSG_TO_SUMMARIZE=10
SUMMARIZATION_LLM_MODEL_PROVIDER_NAME=OpenAI
SUMMARIZATION_LLM_MODEL_NAME=gpt-4o
NGROK_AUTHTOKEN=your-ngrok-authtoken
docker run -p 3978:3978 -e LLM_MODEL_PROVIDER_NAME=OpenAI -e LLM_MODEL_NAME=gpt-4o -e OPENAI_API_KEY=your-key -e AIRFLOW_BASE_URL=http://localhost:8080 -e FERNET_SECRET_KEY=your-secret thedatacarpenter/airflow-copilot:latest
Content type
Image
Digest
sha256:384602684…
Size
287.6 MB
Last updated
about 1 year ago
docker pull thedatacarpenter/airflow-copilot