Row Execution

Row Execution

Each selected row receives a stable ID and an isolated browser execution. One row failure does not terminate unrelated workers.

bun workflow run workflows/signup.json \
  --data=google-sheets://SPREADSHEET_ID/Users \
  --parallel=4 --batch-size=50 \
  --from-row=2 --to-row=500 \
  --where='enabled=true'

--where supports =, ==, !=, >, <, >=, <=, and ~ (case-insensitive contains). Parallelism is bounded, provider reads are incremental, and write-back is serialized in batches.

Temporary browser workers are isolated automatically. A named or explicit browser profile cannot be shared safely by concurrent Chrome processes, so use --parallel=1 with --profile or --user-data-dir.

Use --dry-run to authenticate, inspect, validate, filter, and preview without launching a browser or changing provider data. It still writes a redacted local run summary beneath output/.

The runner limits --parallel to 1–100 workers. --batch-size controls provider reads and queued write-back batches; it does not change browser isolation.

Each completed command writes output/workflow-<run-id>-summary.json. Resumable row state is keyed by the workflow definition and stored separately as output/.automation-state-<workflow-id>.json.