orch

Templates

data-pipeline

Python ETL — DuckDB warehouse + APScheduler jobs. Read/transform/load with cheap local infra.

Tasks
4
Phases
4
Estimated
2.3 h
Routed to
claude/claude-sonnet-4-6
$ orch init --template data-pipeline

The plan

A task starts only when everything above it is done. The violet trace is the longest chain: it sets how soon the plan can finish.

F0.T1 — Python package + DuckDB warehouse fileF0.T1Python package + DuckDBwarehouse fileF1.T1 — HTTP extractor with per-source configF1.T1HTTP extractor with per-sourceconfigF2.T1 — Load raw + transform into martsF2.T1Load raw + transform intomartsF3.T1 — APScheduler entry point + cron configF3.T1APScheduler entry point + cronconfig

Tasks by phase

F0 — Foundation (DuckDB + package)

IDTaskModelEstimateWaits for
F0.T1Python package + DuckDB warehouse fileclaude/claude-sonnet-4-60.4 h—

F1 — Extract

IDTaskModelEstimateWaits for
F1.T1HTTP extractor with per-source configclaude/claude-sonnet-4-60.6 hF0.T1

F2 — Transform + Load

IDTaskModelEstimateWaits for
F2.T1Load raw + transform into martsclaude/claude-sonnet-4-60.8 hF1.T1

F3 — Schedule

IDTaskModelEstimateWaits for
F3.T1APScheduler entry point + cron configclaude/claude-sonnet-4-60.5 hF2.T1

Try your own spec in the playground