SubmitOptions

Checkpoint for when creating a new run.

One or more commands to invoke the graph with.

The assistant config.

The assistant context.

Whether to checkpoint during the run (or only at the end/interruption).

• "async": Save checkpoint asynchronously while the next step executes (default).
• "sync": Save checkpoint synchronously before the next step starts.
• "exit": Save checkpoint only when the graph exits.

Interrupt execution after leaving these nodes.

Interrupt execution before entering these nodes.

The assistant metadata.

Strategy to handle concurrent runs on the same thread. Only relevant if there is a pending/inflight run on the same thread. One of:

• "reject": Reject the new run.
• "interrupt": Interrupt the current run, keeping steps completed until now, and start a new one.
• "rollback": Cancel and delete the existing run, rolling back the thread to the state before it had started, then start the new run.
• "enqueue": Queue up the new run to start after the current run finishes.

Behavior to handle run completion. Only relevant if there is a pending/inflight run on the same thread. One of:

• "complete": Complete the run.
• "continue": Continue the run.

Behavior to handle disconnection. Only relevant if there is a pending/inflight run on the same thread. One of:

• "cancel": Cancel the run.
• "continue": Continue the run.

Callback that is called when an error occurs.

Mark the stream as resumable. All events emitted during the run will be temporarily persisted in order to be re-emitted if the stream is re-joined.

Whether or not to stream the nodes of any subgraphs called by the assistant.

The ID of the thread to fetch history and current values from.