Pausing the Agent for Input

Pause an agent run, hand control to a custom Vue component, and resume with the user's answer.

What is this?

Some agent steps need a person's answer before the run can continue, such as approving a payment or picking a meeting time. useInterrupt pauses the run, shows your Vue component, and resumes the run with the answer. How the pause works depends on your agent framework.

LangGraph ships a first-class interrupt() primitive that lets a running node suspend itself and hand control to the client. The run is frozen server-side until the client resolves the interrupt with a payload, at which point the node resumes as if interrupt() had returned that payload.

CopilotKit's useInterrupt is the frontend half of that contract: it subscribes to the paused run, exposes reactive interrupt state, and resumes the agent with the user's answer.

When should I use this?

Reach for useInterrupt when the pause is a graph-enforced checkpoint where the code path must stop and wait for a human, not an LLM-initiated tool call. Typical cases:

  • A sensitive action (payments, irreversible writes) must be approved
  • A required piece of state isn't known and can only be collected from the user
  • The agent explicitly reaches an approval node in a longer workflow
  • You want the server-side contract to be interrupt(...) and resume with a payload

For LLM-initiated pauses where the model decides on the fly to ask the user, prefer useHumanInTheLoop.

The backend: interrupt() inside a tool

Install the CrewAI AG-UI package

pip install ag-ui-crewai litellm

Give useHumanInTheLoop tools to the model

A Flow makes its own model call, so it must pass the tools itself:

  • Type the Flow on CopilotKitState. A tool registered with useHumanInTheLoop arrives in self.state.copilotkit.actions.
  • Pass the actions as tools, inside copilotkit_stream, so the tool call reaches the browser as it streams.
  • The run ends when the model calls the tool. After the user answers, CopilotKit runs the Flow again with the answer as the last message, with role: "tool". Read it in the @start() step to decide what happens next.
agent.py
from typing import Any, cast

from crewai.flow.flow import Flow, listen, start
from litellm import acompletion

from ag_ui_crewai import CopilotKitState, copilotkit_stream


class AgentState(CopilotKitState):
    pass


class SampleAgentFlow(Flow[AgentState]):
    @start()
    async def check_for_user_feedback(self):
        if not self.state.messages:
            return

        last_message = cast(Any, self.state.messages[-1])

        # The result of the useHumanInTheLoop tool (SEND or CANCEL)
        if last_message["role"] == "tool":
            if last_message.get("content") == "SEND":
                self.state.messages.append({
                    "role": "assistant",
                    "content": "✅ Great! Sending your essay via email.",
                })
                return

            if last_message.get("content") == "CANCEL":
                self.state.messages.append({
                    "role": "assistant",
                    "content": "❌ Okay, we can improve the draft. What would you like to change?",
                })
                return

    @listen(check_for_user_feedback)
    async def chat(self):
        system_message = {
            "role": "system",
            "content": (
                "You write essays. Use your tools to write an essay; "
                "don't just write it in plain text."
            ),
        }

        response = await copilotkit_stream(
            await acompletion(
                reasoning_effort="none",
                allowed_openai_params=["reasoning_effort"],
                model="openai/gpt-6-luna",
                messages=[system_message, *self.state.messages],
                tools=self.state.copilotkit.actions,
                stream=True,
            )
        )

        self.state.messages.append(response.choices[0].message)

Pause the Flow for useInterrupt

To pause at a fixed step in your code instead, decorate the step with CrewAI's @human_feedback and the AG-UI feedback provider. The step's return value is the interrupt payload. The listening step receives the user's answer as a HumanFeedbackResult.

src/agents/interrupt_flow.py
import json
from typing import Any

from crewai.flow import Flow, HumanFeedbackResult, human_feedback, listen, start

from ag_ui_crewai import CopilotKitState, agui_feedback_provider


class InterruptFlow(Flow[CopilotKitState]):
    @start()
    @human_feedback(
        message="Choose a meeting time or cancel the request.",
        llm=None,
        provider=agui_feedback_provider,
    )
    async def request_schedule(self) -> dict[str, Any]:
        # ... call the model, then return what the picker needs:
        return {"topic": topic, "attendee": attendee, "slots": slots}

    @listen("request_schedule")
    async def confirm_schedule(self, result: HumanFeedbackResult) -> None:
        feedback_text = result.feedback or ""
        # A cancelled interrupt resumes with an empty feedback string.
        cancelled = not feedback_text.strip()
        feedback = {} if cancelled else json.loads(feedback_text)

Register the endpoint so it emits standard AG-UI interrupts:

src/agent_server.py
add_crewai_flow_fastapi_endpoint(
    app,
    interrupt_flow,
    "/interrupt",
    emit_interrupt_outcome=True,
    enable_legacy_on_interrupt_event=False,
)

For a Conversational Flow, pass these two flags in addition to conversational=True. On the frontend, read the payload from interrupt.metadata.crewai.output. See src/agents/interrupt_flow.py for the full Flow.

The agent can expose a schedule_meeting tool that issues an interrupt with the meeting context. The run stops until the client resolves the interrupt. The resolve payload becomes the return value of interrupt(). The tool turns that value into a result for the model.

Two things to note:

  • The payload, for example { "topic": topic, "attendee": attendee }, is what the Vue frontend receives as event.value. The LangGraph adapter sends it as a JSON string, so parse it before you read fields. Keep it a plain, serializable object containing the pause-time context the UI needs.
  • The return-side contract, such as { chosen_label, chosen_time } or { cancelled: true }, is entirely yours. The client can send any serializable resolve payload; the tool gives it meaning and must validate it before a side effect.

The frontend: useInterrupt and the chat slot

Register useInterrupt in a descendant component. With the default renderInChat: true, the composable publishes the pending interrupt to CopilotChat; render it through the #interrupt scoped slot and call resolve(...) to resume:

src/components/MeetingChat.vue
<script setup lang="ts">
import { CopilotChat, useInterrupt } from "@copilotkit/vue/v2";

type MeetingPayload = { topic?: string; attendee?: string };

// The LangGraph adapter sends the interrupt value as a JSON string.
function readValue(value: unknown): MeetingPayload {
  if (typeof value !== "string") return (value ?? {}) as MeetingPayload;
  try {
    return JSON.parse(value) as MeetingPayload;
  } catch {
    return {};
  }
}

useInterrupt({
  agentId: "gen-ui-interrupt",
});
</script>

<template>
  <CopilotChat agent-id="gen-ui-interrupt">
    <template #interrupt="{ event, resolve }">
      <section>
        <p>Pick a time for {{ readValue(event.value).topic ?? "a call" }}</p>
        <button
          @click="resolve({ chosen_label: 'Tomorrow', chosen_time: '2026-09-01T09:00:00Z' })"
        >
          Tomorrow at 09:00
        </button>
        <button @click="resolve({ cancelled: true })">Cancel</button>
      </section>
    </template>
  </CopilotChat>
</template>

Whatever you pass to resolve is round-tripped back to the agent as the return value of the matching interrupt.

Ask the agent to book a call. The picker appears in the chat. Click Tomorrow at 09:00, and the agent confirms the booked time.

Key props

  • agentId: must match a runtime-registered agent. If omitted, the composable uses the agent of the surrounding chat configuration, then "default". A mismatch means the interrupt never appears.
  • renderInChat: when true, slotProps are published to the chat's #interrupt slot. Set it to false when your own component renders the returned refs elsewhere. See Headless interrupts.
  • handler: optionally preprocesses the interrupt, can resolve it early, and exposes its return value as result.
  • enabled: filters interrupt shapes before they are surfaced.

The Vue #interrupt slot receives { event, result, resolve }. For the complete standard-interrupt surface (including interrupt, interrupts, and cancel), use the composable's returned slotProps with renderInChat: false.

Multiple interrupts? Add a type and gate with enabled

If your graph issues more than one kind of interrupt, tag each payload with a type field and install one useInterrupt per shape, each gated by enabled:

// The LangGraph adapter sends the interrupt value as a JSON string.
function readValue(value: unknown): Record<string, unknown> {
  if (typeof value !== "string") return (value ?? {}) as Record<string, unknown>;
  try {
    return JSON.parse(value) as Record<string, unknown>;
  } catch {
    return {};
  }
}

useInterrupt({
  agentId: "gen-ui-interrupt",
  enabled: (event) => readValue(event.value).type === "ask",
});

useInterrupt({
  agentId: "gen-ui-interrupt",
  enabled: (event) => readValue(event.value).type === "approval",
});

Preprocess with handler

For cases where the interrupt can sometimes be resolved without user input, pass a handler that runs before the UI is rendered. The handler can call resolve(...) itself to resume the agent early, or return a value exposed as result:

const { slotProps } = useInterrupt({
  agentId: "gen-ui-interrupt",
  renderInChat: false,
  handler: async ({ event, resolve }) => {
    const department = await lookupUserDepartment();
    // readValue is the helper from the previous sample.
    const required = readValue(event.value).accessDepartment;

    if (required === department || department === "admin") {
      await resolve({ code: "AUTH_BY_DEPARTMENT" });
      return;
    }

    return { department };
  },
});

AG-UI standard interrupt flow vs. legacy

This section is a reference for developers who write interrupt UI for their own agent. The Vue useInterrupt reference lists every option. useInterrupt handles two kinds of interrupt, and your agent framework decides which kind it sends.

Standard flow (RUN_FINISHED with outcome.type === "interrupt")

When the agent backend conforms to the AG-UI protocol, it signals an interrupt by emitting a RUN_FINISHED event whose outcome carries the interrupts array:

outcome.type === "interrupt"
outcome.interrupts  // Interrupt[]

The composable detects this on onRunFinishedEvent and exposes the interrupts after onRunFinalized fires. slotProps.value contains:

  • interrupt: the primary Interrupt (interrupts[0]), with shape { id, reason, message?, toolCallId?, responseSchema?, expiresAt?, metadata? }.
  • interrupts: the full open set.
  • resolve(payload?, interruptId?): records a resolved response for the targeted interrupt. The run resumes once every open interrupt has a response.
  • cancel(interruptId?): records a cancelled response for the targeted interrupt. The same accumulate-then-submit logic applies.

Legacy flow (on_interrupt custom event)

Agents can emit a custom on_interrupt event. The LangGraph adapter does this by default. The composable exposes its payload in event.value, with interrupt set to null and interrupts to []. Calling resolve(payload) resumes via forwardedProps.command. cancel() dismisses the interrupt without resuming because the custom-event flow has no cancel response.

Priority

If both signals appear on the same run, the standard flow wins.

Approve / Cancel example

src/components/ApprovalPanel.vue
<script setup lang="ts">
import { useInterrupt } from "@copilotkit/vue/v2";

const { slotProps } = useInterrupt({ renderInChat: false });
</script>

<template>
  <section v-if="slotProps">
    <p>{{ slotProps.interrupt?.message ?? "Approve this action?" }}</p>
    <button @click="slotProps.resolve({ approved: true })">Approve</button>
    <button @click="slotProps.cancel()">Cancel</button>
  </section>
</template>

Both methods return the RunAgentResult once the run restarts, or void while waiting for responses to other open interrupts.

Multi-interrupt behavior

Some agents issue more than one interrupt in a single run. Each interrupt has its own id; address them individually:

<template>
  <ul v-if="slotProps">
    <li v-for="item in slotProps.interrupts" :key="item.id">
      {{ item.message }}
      <button @click="slotProps.resolve({ ok: true }, item.id)">Approve</button>
      <button @click="slotProps.cancel(item.id)">Cancel</button>
    </li>
  </ul>
</template>

The agent run resumes only after every open interrupt has been addressed. If you omit interruptId, the primary interrupt is targeted; Vue warns when several are open.

responseSchema: surface only, no client-side validation

The Interrupt type exposes a responseSchema JSON Schema that the agent can use to describe the expected payload. useInterrupt surfaces it on interrupt.responseSchema for your UI to read, but it does not validate resolve payloads against it. Validation is the agent's responsibility on resume.

Provider setup

The component calling useInterrupt must be a descendant of CopilotKitProvider:

src/App.vue
<script setup lang="ts">
import { CopilotKitProvider } from "@copilotkit/vue/v2";
import MeetingChat from "./components/MeetingChat.vue";
</script>

<template>
  <CopilotKitProvider runtime-url="/api/copilotkit">
    <MeetingChat />
  </CopilotKitProvider>
</template>

How is this guide?