InvocationsHostServer

The underlying :class:InvocationAgentServerHost.

The hosted runnable.

Start the server synchronously.

Once running, the host exposes POST /invocations. The default request body is:

.. code-block:: json

{"message": "Hello!", "stream": false}

message is either user text or a list containing a structured HITL response item. stream is optional and defaults to false. Non-streaming requests return JSON:

.. code-block:: json

{"response": "Assistant text"}

A pending LangGraph interrupt adds Responses-style function_call and mcp_approval_request items under output. Resume it by sending the matching function_call_output or mcp_approval_response as the next request's message list.

Streaming requests return text/event-stream with token payloads:

.. code-block:: text

data: {"token": "..."}

event: done
data: {}

With options.resilient_background=True, a request containing {"message": "Hello!", "background": true} returns 202. Retrieve or cancel it through GET /invocations/{invocation_id} and POST /invocations/{invocation_id}/cancel respectively.

Multi-turn callers should reuse the x-agent-session-id response header as the next request's agent_session_id query parameter.

Start the server asynchronously.

Exposes the same POST /invocations contract as :meth:run. The default request body is:

.. code-block:: json

{"message": "Hello!", "stream": false}

Non-streaming requests return {"response": "Assistant text"}. Streaming requests return text/event-stream with data: {"token": "..."} payloads, any pending HITL items as event: output_item, and finally event: done. Resilient background requests return 202 and are observed through the invocation GET and cancel endpoints. Multi-turn callers should reuse the x-agent-session-id response header as the next request's agent_session_id query parameter.

Parse the invocation request body.

Default implementation reads message as either non-empty text or a non-empty list of structured HITL response items, plus an optional boolean stream. Override to support a different body schema.

Parse task-backed execution options from the request body.

Override this hook alongside :meth:parse_request when a custom body schema carries background or continuation controls elsewhere.

Build a RunnableConfig for the invocation.

Sets configurable.thread_id from a user-scoped internal form of request.state.session_id so checkpointed conversations cannot collide across users that choose the same public session ID.

Build config for a task-backed invocation attempt.

The graph thread uses context.task_id, the user-partitioned internal ID for the public session. The public session_id remains available in the invocation envelope and task input only.

The task context is transport-scoped rather than checkpointed graph state. Nodes can use it to inspect recovery, cancellation, steering, and shutdown signals for the current attempt.

Build the runnable input from the parsed message.

Default implementation produces {"messages": [HumanMessage(...)]}. Override to support custom-state graphs.

Translate a non-streaming runnable result into response text.