majiayu000 Babysitter Orchestration
Use this skill to orchestrate complex, multi-step coding workflows through babysitter's process-driven execution engine.
How It Works
- Babysitter creates a run from a process definition (JavaScript file with tasks)
- Each iteration produces effects (tasks to execute)
- You execute the effects and post results back
- The loop continues until the run completes
Starting a Run
Use the SDK bridge directly (imported from the extension):
createNewRun()to create a runiterate()to advance the orchestrationpostResult()to report effect outcomes
Effect Types
| Kind | Action |
|---|---|
| agent | Execute as sub-agent task |
| node | Run Node.js script |
| shell | Run shell command |
| breakpoint | Ask user for approval |
| sleep | Wait until timestamp |
| skill | Expand and execute skill |
Completion
When the run completes, the SDK returns a completion proof. Output it wrapped in <promise>PROOF_VALUE</promise> tags.