Latest alpha · 0.1.0a4

Make a growing script understandable before you run it.

kazeflow turns ordinary Python functions into a small, inspectable flow. Review the dependency graph, explicitly approve execution, and keep a useful result—without a service, daemon, or required runtime dependency.

Run your first flow View on GitHub

• Start in one file

  Install kazeflow, declare two assets, inspect the plan, and deliberately run it.

  Getting started

• Review the work first

  Inspect targets and dependencies in text, JSON, Mermaid, or DOT.

  CLI and graph output

• Rerun one slice

  Select a date, region, file, or another script-defined partition key.

  Partition guide

• Look up an exact contract

  Find command options, exit statuses, and public Python signatures.

  Reference

Where kazeflow fits

Start with a script. When it grows into several dependent steps, keep the functions plain Python and add enough structure to see the work order, choose targets, and inspect the outcome. kazeflow is intentionally not a scheduler, remote-worker system, database-backed platform, or sandbox.

Trust the Python entry before loading it

Every CLI command imports the supplied Python entry. Import-time code and an explicitly selected factory can run before kazeflow discovers the flow. plan does not invoke decorated asset bodies after loading; it does not make an untrusted script safe. Understand the trust boundary.

Choose how to read

You want to…	Start here
Run kazeflow for the first time	Getting started
Complete a specific task	Guides
Understand the execution model	Concepts
Check an exact command or signature	CLI reference or Python API
Adapt a complete pattern	Examples