Daily rollups without an orchestrator

You have raw event data piling up in a sqlite URL. Every night you want to compute the daily rollup table: counts by hour, top users, error rates.

The whole pipeline is one hoody-cron entry. At 1am it curls a hoody-exec endpoint. The exec script runs the SQL, writes results back to the same sqlite URL, exits. No DAG, no Airflow Postgres, no scheduler dashboard with 14 widgets, no on-call rotation for the orchestrator itself.

If the rollup fails, the cron logs say so. If you need to backfill, you curl the exec URL by hand with a date parameter. The whole pipeline is two URLs and a 5-field schedule.

The orchestrator is a cron entry pointing at a URL.