Blueprint

Turns Airflow DAG authoring into a template system where you write validated Python blueprints once, then compose them into DAGs using YAML files. The setup is straightforward: define a Blueprint class with Pydantic validation, implement a render method that returns a TaskGroup, then reference it in YAML with step-specific config. Supports Jinja2 templating with access to Airflow variables and runtime context, plus optional runtime parameter overrides through the UI. The real value is for teams that want to give non-engineers a safe, linted interface to create DAGs without touching Python, or when you're building the same patterns repeatedly and want enforcement around configuration. Requires Python 3.10+ and Airflow 2.5+, which might rule it out for older deployments.

Install to Claude Code

npx -y skills add astronomer/agents --skill blueprint --agent claude-code