Executions

Each run of a task, with logs, events, pending items and text search.

📄️List executions

List executions

📄️Get execution details

Get execution details

📄️Cancel the execution

Cancel the execution

📄️Continue the conversation (next round)

Only for a conversational Task (`conversational`): creates a child execution of this one that resumes the agent session (context, output files and state) and handles the request in `message`. It uses this execution's parameters, reference date and environment, and this execution must be finished (`success` or `failed`). Continuing an older round opens a branch. The new execution carries `parentExecutionId` (this one) and `rootExecutionId` (the first round).

📄️Text search across executions

Searches events, todos, logs, result and error. Up to 30 executions, most recent first, each with up to 3 snippets. Only sources the caller could read directly are included: logs require the organization's `executions_logs.view` permission, and on a person's token the role needs `agent.logs.list`, `agent.events.list` and `agent.todos.list` for logs, events and todos.

📄️Execution logs

In order of arrival, up to 1000 per call. To follow a running execution, pass the last `id` received in `after_id` and repeat.

📄️Execution events

By `sequence`, up to 500 per call. To follow along, pass the last `sequence` received in `after_sequence` and repeat.

📄️Pending items (todos) of the execution

Pending items (todos) of the execution