2026-07-18 · Updated 2026-07-19 · 1 min read

Choose Workflow Runner or Parallel Runner

Match ordered handoffs and independent fan-out to the runner that owns their failure and artifact model.

By Juno AI INC · workflow-runner · parallel-runner

The decision is about data dependency, not the number of commands. Ask whether any item needs another item’s response, artifact, or session before it can start.

Choose Parallel Runner for fan-out

Use Parallel Runner when tasks are independent and a bounded worker pool can process them safely.

sh
./.juno_task/scripts/parallel_runner.sh --kanban-filter "ready" --parallel 3

Its evidence is item-oriented: per-task JSON, status, logs, and aggregation. Raw command-file mode can fan out several complete workflow files.

Choose Workflow Runner for flow

Use Workflow Runner when ordered steps exchange steps.<id>.response, files, or captured sessions.

sh
./.juno_task/scripts/workflow_runner.sh lint --workflow release-review.yaml
./.juno_task/scripts/workflow_runner.sh --workflow release-review.yaml --dry-run

Its evidence is run-oriented: manifest, rendered workflow, step streams and responses, summary, and handoff session IDs.

Compose them deliberately

A workflow can launch a controlled parallel investigation, then consume its aggregation artifact in a review step. Parallel command mode can also run several self-contained workflows. Keep one owner for each ordering boundary and never represent a real dependency as timing luck.