Human-in-the-Loop

  Start your AG2 backend on a `/chat` endpoint and connect your frontend to it with CopilotKit.
</Step>

<Step>
    ### Create a frontend human-in-the-loop tool

    Frontend tools can be leveraged in a variety of ways. One of those ways is to have a human-in-the-loop flow where the response
    of the tool is gated by a user's decision.

    In this example we will simulate an "approval" flow for executing a command. First, use the `useHumanInTheLoop` hook to create a tool that
    prompts the user for approval.

    ```tsx title="page.tsx"

    export function Page() {
      // ...

      useHumanInTheLoop({
        name: "offerOptions",
        description: "Give the user a choice between two options and have them select one.",
        parameters: [
          {
            name: "option_1",
            type: "string",
            description: "The first option",
            required: true,
          },
          {
            name: "option_2",
            type: "string",
            description: "The second option",
            required: true,
          },
        ],
        render: ({ args, respond }) => {
          if (!respond) return <></>;
          return (
            <div>
              <button onClick={() => respond(`${args.option_1} was selected`)}>{args.option_1}</button>
              <button onClick={() => respond(`${args.option_2} was selected`)}>{args.option_2}</button>
            </div>
          );
        },
      });

      // ...
    }
    ```
</Step>
<Step>
    ###  Set up your AG2 backend

    The frontend tool emits AG-UI tool events. AG2 can consume those events, then persist the user decision to shared state:

    ```python title="agent.py"
    from typing import Annotated

    from ag_ui.core import EventType, StateSnapshotEvent
    from fastapi import FastAPI, Header
    from fastapi.responses import StreamingResponse
    from autogen import ContextVariables, ConversableAgent, LLMConfig
    from autogen.ag_ui import AGUIStream, RunAgentInput

    agent = ConversableAgent(
        name="assistant",
        system_message=(
            "When you need the user to choose, call the frontend tool `offerOptions`. "
            "After it returns, call `store_user_choice` with the selected value."
        ),
        llm_config=LLMConfig({"model": "gpt-5.4-mini"}),
    )

    @agent.register_for_llm(description="Store the latest user choice in shared state.")
    def store_user_choice(
        context: ContextVariables,
        choice: Annotated[str, "The option selected by the user"],
    ) -> StateSnapshotEvent:
        snapshot = {"hitl": {"latest_choice": choice}}
        context["hitl"] = snapshot["hitl"]
        return StateSnapshotEvent(type=EventType.STATE_SNAPSHOT, snapshot=snapshot)

    agent.register_for_execution(name="store_user_choice")(store_user_choice)

    stream = AGUIStream(agent)
    app = FastAPI()

    @app.post("/chat")
    async def run_agent(
        message: RunAgentInput,
        accept: str | None = Header(None),
    ):
        return StreamingResponse(
            stream.dispatch(message, accept=accept),
            media_type=accept or "text/event-stream",
        )
    ```

    The frontend tools are automatically populated by CopilotKit through the AG-UI protocol and are available to your AG2 backend.
</Step>
<Step>
    ### Try it out

    You've now given your agent the ability to show the user two options and have them select one. The agent will then be aware of the user's choice and can use it in subsequent steps.

    ```
    Can you show me two good options for a restaurant name?
    ```
</Step>