Skip to main content
A Task request can finish immediately, wait in the execution queue, or continue while your integration receives updates. Use the execution endpoints to build dashboards, polling workers, and operational tooling.

​
Understand execution state

The lightweight execution list reports the state of each run: Use GET /api/executions/list for a summary view and GET /api/executions/:id when you need logs and the result payload.

​
Poll for completion

A simple worker can list recent executions and fetch the detail for a specific execution:
Use polling when your integration does not support server-sent events. Stop polling after your own application timeout.

​
Stream live updates

The REST API exposes a server-sent events endpoint at GET /api/executions/stream for clients that need live execution updates. Use an SSE-compatible client to keep a connection open and process events as they arrive. This is useful for an operations dashboard or an interface that shows a Task’s progress without repeatedly requesting the full execution record.

​
Retrieve a full result

Large persisted results can be trimmed from an execution history record. When that happens, the record is marked with truncated: true and hasFullResult: true. Request the complete result with:
The full result is stored only while the execution record is retained. See Captures and Storage for retention behavior.

​
Stop an execution

Stop a running execution or cancel one that is still queued:
A queued run that is cancelled never starts. The request returns 400 MISSING_RUN_ID when runId is missing.

​
Handle capacity and timeout responses

Execution requests can fail when resource protection is active: For 503 responses, use the retryAfterMs response value and Retry-After header to decide when to retry. See REST API Reference for the complete execution endpoint details.