LangGraph · stateful workflow webhooks

Stateful workflows without timeout anxiety.

LangGraph Platform pings your endpoint when a node resumes. The node might fire two more LLM calls before it's done. AnyHook gives you the headroom.

Start free — 3K events / mo How AnyHook works

The timing problem

LangGraph Platform's webhook callback signals that a paused workflow is ready to resume. The resumed node often triggers further model invocations — that's the whole point of stateful workflows.

If you're chaining LangGraph into a Stripe or GitHub event, you're stuck: the upstream sender has 10 seconds, LangGraph and its downstream calls need 60 to 120. AnyHook decouples the two timelines.

60–120s

Typical LangGraph resume latency

10s

Upstream sender SLA

AES-256-GCM

Payload encrypted at rest

One-click

Replay any past event

Forward LangGraph callbacks via AnyHook

Configure LangGraph to call AnyHook's inbound URL when the workflow needs to resume against your handler. AnyHook ack-and-forwards, you process at your own pace.

import { verifyWebhook } from "@anyhook/verify";
import { Client } from "@langchain/langgraph-sdk";

const lg = new Client({ apiUrl: process.env.LANGGRAPH_API_URL });

export async function POST(req: Request) {
  if (!(await verifyWebhook(req, process.env.ANYHOOK_SIGNING_SECRET!))) {
    return new Response("invalid", { status: 401 });
  }

  const { thread_id, node_id, payload } = await req.json();

  // Resume the LangGraph thread. Two more model calls + a vector
  // search are common here. Total ~90s.
  const result = await lg.runs.create(thread_id, "agent", {
    input: payload,
    config: { configurable: { resume_from: node_id } },
  });

  await persistRun(result);
  return new Response("ok", { status: 200 });
}

Questions teams ask

Does AnyHook handle LangGraph's resumable-state semantics?: AnyHook is HTTP-layer infrastructure — it doesn't model LangGraph's checkpoints. Your handler still owns workflow state. AnyHook just guarantees the resume signal reaches your handler reliably, with retries and replay.
Can I fan out one LangGraph callback to multiple downstream services?: Yes. Configure multiple destinations on the AnyHook app and each delivery is independent — one can succeed while another retries.
How long can a single delivery attempt run?: 60 seconds on Free, 120s on Pro, 300s (5 min) on Scale. Most LangGraph resume callbacks fit comfortably in the Pro window.

Change one URL. Keep your LangGraph handler.

Free tier covers 3K events / month. No SDK, no code changes — just point LangGraph at your AnyHook inbound URL.

Start free See pricing