airflow-dag-design

name	airflow-dag-design
description	Design small, maintainable Airflow DAGs with clear task boundaries, retry policy, and runtime assumptions.
license	MIT
compatibility	opencode
metadata	{"audience":"data-engineers","domain":"orchestration"}

What I do

• Break a pipeline into task-sized Airflow steps.
• Recommend schedules, retries, ownership, and tags.
• Keep shared logic in src/ and DAG wiring in dags/.
• Highlight idempotency and backfill concerns.

Design rules

• DAG files should stay thin.
• Business logic should not be embedded directly inside operators when it can live in shared Python modules.
• Document retries, timeout assumptions, and expected outputs.