neochaotic/leoflow — reverse-engineered prompt

Reverse engineered prompt

GitHub

Build me a pre alpha version of Leoflow, a fast Go workflow orchestrator for teams that like Airflow but want something simpler and more container native.

I want a CLI where someone can start a DAG project, validate it, compile a small Python DAG into a versioned JSON artifact, push it to a control plane, and create auth tokens. The server should expose Airflow compatible API endpoints, use JWT auth with basic roles and tenants, store metadata in Postgres, schedule runs from cron, and move DAG runs and task instances through a deterministic state machine until tasks are queued.

Please include Prometheus metrics, structured logs, OpenTelemetry hooks, database migrations, local Docker compose for development, and clear docs for trying it locally. It’s okay if real task execution, logs, XCom, and the actual Airflow UI are not finished yet, but leave clean interfaces for the executor and agent so those can be added next. Look up current docs online if you need to.

Want more depth? Deep Reverse