pg-task-tracker

A shoebox, not a warehouse. Celery, Airflow, and Dramatiq are excellent if you need to execute tasks across workers — but they’re heavy machinery if all you want is a durable record of what happened during a multi-step process. pg-task-tracker is the lighter option: two tables in your existing Postgres (or SQLite) database, an ORM-friendly API built on SQLModel/SQLAlchemy, and nothing else to install or run.

The core shape is tasks → steps, each with a UUID, a status, and an arbitrary JSON metadata payload. Use the manual API to create tasks and update steps as work progresses, or wrap a function with the @track decorator and let the library handle status transitions for you. A raw SQL migration is included for projects that manage their own schema; an ensure_schema() call covers the rest.

I wrote about why I built it in Tracking multi-step Python tasks with Postgres.