Executions

​

How executions start

• Manual — Run from Properties → Execute.
• API/Schedule — Triggered externally (for exposed top‑level Nodes).
• Internal — Invoked by another Node within a Flow.

​

Characteristics

• Isolated — No shared state between runs.
• Reproducible — Inputs, modifiers, and context are logged.
• Traceable — Inspect logs, traces, and metrics historically.
• Evaluated — Optionally score outputs with Evaluations.

​

Flow of an execution

1. Trigger — Manual, schedule, or API.
2. Input — JSON payload (from trigger or saved Payload).
3. Process — Node(s) execute and pass data along Edges.
4. Output — Result returned or stored.

​

Practical tips

• Save Payloads for common scenarios.
• Keep inputs small and typed for easier regression.
• Use consistent correlation fields to stitch traces across components.

note: Parallelism — For components with Parallel ON, provide a list of inputs; each item is executed concurrently and aggregated downstream.