HITL Overview

Allow your agent and users to collaborate on complex tasks.

Not available for Langchain DeepAgents yet

The code on this page comes from the showcase's hitl-in-chat demo, and Langchain DeepAgents doesn't have it yet. See this page for CopilotKit's Built-in Agent, LangGraph (Python), LangGraph (TypeScript).
See this in Inspector

Open Inspector on localhost. Go to Inspect, then Frontend Tools. Your tool and its schema are listed.

More detail: Inspector.

What is this?

Human-in-the-loop (HITL) lets an agent pause mid-run to collect input, confirmation, or a choice from the user, then resume with that answer folded back into its reasoning. It's what turns an autonomous workflow into a collaborative one: the agent keeps its context, the user keeps the steering wheel.

When should I use this?

Use HITL when you need:

  • Quality control: a human gate at high-stakes decision points.
  • Missing input: the agent needs a value only the user knows, such as a meeting time.
  • Risky actions: the agent must get a yes before it sends, pays, or deletes.
  • Expert input: the user supplies domain knowledge the model lacks.

Two patterns for HITL in CopilotKit

CopilotKit ships two complementary ways to pause an agent turn and ask the human something. They look similar from the outside (the chat pauses, a custom component appears, the user answers, the run resumes) but they're wired differently on the backend, and each has its own niche.

PatternWho decides to pause?Backend surface
useHumanInTheLoopThe LLM, by calling a registered client-side toolA frontend-only tool description (Zod schema + render)
useInterruptYour agent code, at a step that must stop for a personAn interrupt your agent emits, for example LangGraph interrupt() or Mastra suspend()

Pick useHumanInTheLoop when the pause is an agent-initiated decision (the model chose to ask the user) and you want the picker UI inlined into the normal tool-call flow.

Pick useInterrupt when the pause is a checkpoint in your agent code: the code path always requires a human answer, and the server-side contract is an interrupt. On the Built-in Agent, interrupts are temporarily unavailable, so use useHumanInTheLoop there.

Deep Agents compile to LangGraph graphs, so they support both patterns. CopilotKitMiddleware forwards a useHumanInTheLoop tool to the model, the same way it forwards any frontend tool. For useInterrupt, call LangGraph's native interrupt() from a middleware hook. Deep Agents has no live demo on these pages yet, so the steps below carry the code.

Call interrupt from a middleware hook

A Deep Agent has no custom nodes, so put the pause in a middleware hook. This hook runs before the model call and asks the user to name the agent the first time. The middleware also declares the agent_name state field that stores the answer.

agent.py
from typing import Any, NotRequired

from copilotkit import CopilotKitMiddleware, CopilotKitState
from deepagents import create_deep_agent
from langchain_openai import ChatOpenAI
from langchain.agents.middleware import AgentMiddleware
from langgraph.runtime import Runtime
from langgraph.types import interrupt

class AgentState(CopilotKitState):
    agent_name: NotRequired[str]

class AgentNameMiddleware(AgentMiddleware[AgentState, Any]):
    state_schema = AgentState

    def before_model(self, state: AgentState, runtime: Runtime[Any]) -> dict[str, Any] | None:
        if not state.get("agent_name"):
            # Interrupt and wait for the user to respond with a name
            name = interrupt("Before we start, what would you like to call me?")
            return {"agent_name": name}
        return None

agent = create_deep_agent(
    model=ChatOpenAI(model="gpt-6-luna", reasoning_effort="none"),
    middleware=[
        AgentNameMiddleware(),
        CopilotKitMiddleware(expose_state=["agent_name"]),
    ],
    system_prompt=(
        "You are a helpful assistant. After the user chooses a name, "
        "Current agent state contains agent_name. Use that value as your own name."
    ),
)

Make the agent aware of the answer

Resuming an interrupt and telling the model about the answer are separate steps. By default the model is not aware of the interaction: the run resumes, and the model never sees what the user chose.

The custom middleware saves the answer in thread state. In TypeScript, zodState also keeps the custom field in the graph's output, so AG-UI can send it to the frontend. The CopilotKit middleware then adds only the allowlisted field (expose_state / exposeState) to each model request, and the system prompt says what that field means. State exposure is off by default, so without the allowlist the model never sees the name. Keep the allowlist narrow, and do not expose unrelated application state.

Handle the interrupt in your frontend

Give useInterrupt a component to render, and call resolve with the user's answer. Whatever you pass to resolve becomes the return value of interrupt().

app/page.tsx
import { useInterrupt } from "@copilotkit/react-core/v2";
// ...

const YourMainContent = () => {
  // ...
  // styles omitted for brevity
  useInterrupt({
    render: ({ event, resolve }) => (
      <div>
        <p>{event.value}</p>
        <form onSubmit={(e) => {
          e.preventDefault();
          resolve((e.target as HTMLFormElement).response.value);
        }}>
          <input type="text" name="response" placeholder="Enter your response" />
          <button type="submit">Submit</button>
        </form>
      </div>
    )
  });
  // ...

  return <div>{/* ... */}</div>
}

The AG-UI LangGraph adapter sends an object interrupt value as a JSON string, so parse event.value before you read fields from it.

When your app already selects an agent in its CopilotKit provider, useInterrupt uses that agent. Pass agentId only when this component must target a different agent in a multi-agent app.

Pattern 1: useHumanInTheLoop (tool-based)

Register a HITL tool in your React app with useHumanInTheLoop. When the LLM calls that tool, CopilotKit routes the call through your render function, which shows a custom component and calls respond with the user's answer. The agent sees the answer as the tool result and continues from there.

The sample imports TimePickerCard from its own file. The card renders the slots and calls onSubmit with the user's choice. These are its props:

Each slot button calls onSubmit({ chosen_time: s.iso, chosen_label: s.label }). The "None of these work" button calls onSubmit({ cancelled: true }). The full file is in the Code view of the demo at the top of this page.

The picker UI is fed a static list of candidate slots. This is just data the demo page owns, so you can swap in real availability, a calendar API, or anything else:

Pattern 2: useInterrupt (agent-paused)

With an interrupt, your agent code enforces the pause, for example with LangGraph's interrupt():

  1. A node calls interrupt({...}), and the run stops.
  2. The client receives the payload and renders your UI.
  3. The user answers, and the run resumes with the answer.

CopilotKit's useInterrupt hook is the render contract.

See the useInterrupt deep dive for the full walkthrough, including the backend tool and render-prop wiring.

Going headless

Both patterns above ship with a render prop. CopilotKit handles the "when to show the picker" logic for you. If you want to drive interrupt resolution from a custom UI that lives anywhere in the tree (not necessarily inside a chat), see the headless interrupts guide. It shows how to compose useAgent, agent.subscribe, and copilotkit.runAgent to build your own useInterrupt equivalent.

How is this guide?