ClawdCall
Log In

Turn an n8n error workflow into a phone escalation

Use an n8n Error Trigger to catch an automatic workflow failure, let ordinary errors retry, and call one trusted owner only when the run is still blocked and a decision is time-sensitive.

Download the n8n workflow JSONTest it with your own number

Independent guide; not affiliated with n8n. Test with your own number first. Calls may use minutes.

01Run stalls
02Check urgency
03Call owner
04Resume or stop

The event

Detect a blocked execution

Retries are exhausted, or an approval is nearing its deadline.

The checks

Confirm a human decision is needed

The job is still blocked, delay matters, an owner is assigned, and this event has not already caused a call.

After the call

Use the owner's answer

n8n receives one answer: retry, approve, reject, delegate, pause, or open incident.

How do you escalate an n8n error workflow by phone?

Link an n8n Error Trigger to the workflows you need to monitor. Let transient node errors use retries or an error output first. If an automatic run is still blocked, use an HTTP Request node to send an allowlisted summary and a trusted recipient to ClawdCall. Match the returned call ID to the execution before n8n retries, pauses, delegates, or opens an incident.

Decision guide

Choose how n8n should handle a failed workflow

Most failures should not create a phone call. Match the response to whether the workflow can recover automatically, needs routine visibility, or is waiting on a time-sensitive human decision.

Best for

Retry or Error Output

Transient or locally recoverable node errors. Retry the operation or route the error to a fallback branch while the current workflow still knows how to recover.

Best for

n8n Error Workflow

Centralized logging and ordinary notifications. Use an Error Trigger to record failed automatic runs and notify Slack, email, or another normal response channel.

Best for

ClawdCall escalation

A blocked run that needs a decision now. Use a guarded HTTP request to call one trusted owner, collect a constrained answer, and return it to n8n.

Test the ClawdCall route with your number

Call rules

Call only when

  • The run is still blocked after the retry limit
  • A real deadline is inside the configured window
  • The assigned owner can act now
  • This run has not already triggered a call

Example message

What they hear

“The payments workflow is still blocked after three attempts. Reply retry, pause, or delegate.”

Setup

Connect the pieces

  1. 01

    Import the workflow JSON.

  2. 02

    Add your ClawdCall API key as an n8n Header Auth credential.

  3. 03

    Set the test number to your own E.164 number.

  4. 04

    Run once with the HTTP node disabled. Check the payload, then enable one test call.

Download the n8n workflow JSON

Code

n8n call rules

const event = $json;
const allowedState = ["error", "waiting"].includes(event.state);
const retriesExhausted = Number(event.failure_count) >= 3;
const deadlineIsClose = Number(event.deadline_minutes) <= 15;
const safeRecipient = event.owner_known === true;
const material = ["customer", "revenue", "operations", "test"].includes(event.impact);
const approvedSyntheticTest = event.synthetic === true && event.self_test === true;

if (!(allowedState && retriesExhausted && deadlineIsClose && safeRecipient && material)) {
  return [];
}

if (event.synthetic && !approvedSyntheticTest) return [];

return [{ json: { ...event, call_allowed: true } }];

First run

Try it without a live event

  1. Use the sample event and your own number.
  2. Check that the payload contains no customer data, credentials, or raw errors.
  3. Place one call and save its call ID and answer.
Test with your own number

Before it goes live

Keep the call focused

  • Call only after retries fail and a deadline is at risk.
  • Keep keys, stack traces, customer data, and prompts out of the call.
  • Allow one call per run and respect quiet hours.

Handling the result

Return the phone outcome to the originating n8n execution

Store the call ID with the n8n run ID, then match the result before the workflow continues.

  1. 01

    Save the run ID and call ID together.

  2. 02

    Authenticate the callback, or retrieve the result using the call ID.

  3. 03

    Accept only the listed answers and record the result once.

n8n workflow FAQ

Common setup and handoff questions

How do I handle errors in n8n?

Use node retries or an error output for failures the workflow can handle locally. Use a separate Error Workflow with an Error Trigger for automatic executions that stop. Add a phone escalation only when the run remains blocked and a human decision is time-sensitive.

What is an n8n Error Workflow?

It is a separate workflow that starts with the Error Trigger and receives failure information from linked automatic workflows. It can log the failure, notify a normal channel, or pass a carefully filtered event to another system.

Can n8n call a phone number when a workflow fails?

Yes. An HTTP Request node can invoke a calling API such as ClawdCall. This recipe waits until retries are exhausted, resolves the recipient from trusted configuration, and sends only an allowlisted summary rather than the raw error.

Does the n8n Error Trigger run during manual tests?

No. n8n's Error Trigger runs when an automatic workflow fails. Use the included Manual Trigger and synthetic fixture to inspect the gate and payload before enabling one self-call.

How do I prevent duplicate phone calls from n8n errors?

Create a durable idempotency record from the execution or approval ID before the HTTP Request node. Reject repeat IDs, store the returned call ID, and do not rely on execution-local memory for production deduplication.

Sources

Documentation used for this recipe