DAG authoring
Author a DAG: leoflow.yaml plus dag.py compiled to one immutable artifact.
Author DAGs
Author DAGs on the Airflow SDK — operators, dbt, variables & connections, alerting, map-reduce, and worked examples.
Everything about writing DAGs for Leoflow. A DAG is a dag.py on the Airflow Task
SDK plus a leoflow.yaml for packaging and bindings, compiled to one immutable
artifact.
New to authoring? Start with DAG authoring for the project layout and compile model, then reach for the guide that matches your task below.
Highlight
dbt projects as DAGs
Render a dbt project into native model-level tasks — pod-per-task, no Airflow and no Cosmos at runtime.
Run dbt as a DAG →
DAG authoring
The project layout, the two files (
dag.py + leoflow.yaml), and the compile model.
Learn the model →
Airflow compatibility
You write standard Airflow Task SDK code; Leoflow adds a thin runtime and leoflow.yaml — it never re-implements Airflow's Python API.
See the model →
Map-reduce for ML
Fan-out + reduce expressed as a Python list comprehension — native, no extra operators.
Fan out & reduce →
Operators & sensors
Use Airflow operators and sensors from your tasks.
Use operators →
Variables & Connections
Expose Variables and Connections to your task pods, delivered by the control plane.
Wire secrets in →
On-failure alerting
Notify on run failure straight from leoflow.yaml — no extra task.
Set up alerts →
The Lite web editor
Edit and run DAGs from the browser — the fastest inner loop for local dev.
Edit in the browser →
Examples
Runnable example DAGs you can copy and adapt.
Browse examples →
ETL case study
A worked 1 GB ETL on the ephemeral per-run staging volume.
Read the case study →Python authoring & Airflow compatibility
How Leoflow relates to the Airflow Python API: you write standard Apache Airflow Task SDK code, and Leoflow adds a thin runtime plus a packaging file — it never re-implements Airflow’s Python surface.
Airflow operators & sensors
Use Airflow operators and sensors from your DAGs on Leoflow.
dbt projects as DAGs
Render a dbt project into a Leoflow DAG with native model-level tasks.
Variables & Connections
Expose Variables and Connections to your task pods.
On-failure alerting
Notify on run failure from leoflow.yaml — Slack or a generic webhook, no extra task and no Python.
Map-reduce for ML
Fan-out plus reduce as a Python list comprehension — native map-reduce for ML/AI.
Examples
Runnable example DAGs covering the common authoring patterns.
Case study: 1 GB ETL on staging
A worked 1 GB ETL that shares data between tasks through the per-run staging volume.
The Lite web editor
Edit and run DAGs from the browser in Leoflow Lite.