# Automation alerts

> Route Zapier, Make and n8n failure alerts into one Slack channel Ray reads, with the setup steps per vendor and what Ray can and cannot do with them.

Source: https://heyray.io/docs/automation-alerts

Zapier, Make and n8n can each post their failures into Slack. Send them to one
channel, invite Ray, and ask it what failed overnight. There is no integration
to connect and nothing to install beyond Ray itself. Vendor steps checked
against their documentation on 4 September 2026.

## 1. One channel for every alert

Create a channel called `#automation-alerts`, or reuse the one your alerts
already land in. Invite Ray the way you would invite a teammate, with
`/invite @Ray`. Give the channel a topic that says what lands there, so the
next person who joins knows too.

- One channel for all three tools. Ray groups by automation when you ask.
- Keep conversation out of it. Fewer messages means a faster read-back.
- Public or private both work. Ray reads what it is invited to and nothing else.

When the first alert lands, ask:

> **You:** @Ray what just failed in #automation-alerts?

## 2. Zapier

Zapier's own app, Zapier Manager, fires when one of your Zaps errors or gets
turned off. Pair it with a Slack step and the alert lands in your channel with a
link back to the Zap. It is two steps, so it runs on Zapier's free plan.

1. Sign in as the person who created the Zaps you want watched. On Team and
   Enterprise plans the error goes to the Zap's creator, and Zapier Manager only
   sees errors on Zaps that account owns.
2. Check that error emails are on. Open Settings, then Notifications, then My
   notification settings, and set your default notification frequency to
   anything other than Never. Zapier Manager only fires for Zaps that would have
   emailed you.
3. Create the alert Zap, or start from Zapier's template "Send Slack
   notifications if your Zaps run into errors"
   (https://zapier.com/apps/slack/integrations/zapier-manager/24131/send-slack-notifications-if-your-zaps-run-into-errors).
   Trigger: Zapier Manager, New Zap Error. Action: Slack, Send Channel Message,
   into #automation-alerts. Put the Zap's title, the edit link and the error
   text in the message, in that order.
4. Create a second Zap for Zap Turned Off, with the same Slack step. Zapier
   switches a Zap off after repeated errors, and a Zap that is off raises no
   more errors.

Worth knowing:

- The error trigger fires only after Zapier has finished its automatic replays,
  so an alert can land a while after the first failure.
- When an error handler catches the error, Zapier sends no email, and Zapier
  Manager sees nothing either.
- In November 2025 Zapier staff confirmed that New Zap Error was not firing
  reliably, with no fix date. Keep the email notifications on as a backstop.

## 3. Make

Make posts errors from inside the scenario, one error route per module, and
emails you when it deactivates a scenario. The first is a Slack message you
design. The second is email only, so it needs a forward.

1. Add an error route. In the scenario editor, right-click the module that can
   fail and choose Add error handler. Start with the modules that call outside
   services. Step 4 catches what the rest miss.
2. Post from the route. Add Slack, Create a Message, into #automation-alerts.
   Use the scenario name, the error message and the execution id from Make's
   variables, so the alert says which scenario, what broke and which run.
3. End the route with a directive, such as Resume or Skip, so the run continues
   and the alert still goes out.
4. Forward the deactivation email. Make disables a scenario's schedule after a
   run of consecutive errors and tells you by email, with no Slack option. A
   three-module scenario forwards it: a Custom mailhook trigger, a filter on the
   subject so only deactivation notices pass, and Slack, Create a Message. The
   same hook receives Make's per-run error emails and digests, which is why the
   filter matters.

Worth knowing:

- With Store incomplete executions switched on, an error becomes a warning and
  the run parks in Incomplete executions. Nothing posts unless your error route
  does.
- Once your error routes post to Slack, keep Deactivation on in Make's email
  preferences and switch Errors and Warnings off. A module with no route then
  only surfaces at deactivation, and the error emails stop.

## 4. n8n

n8n has the cleanest answer of the three: an error workflow that runs whenever
a production workflow fails, carrying the workflow's name, the error message and
a link to the failed execution.

1. Create a workflow that starts with the Error Trigger node
   (https://docs.n8n.io/integrations/builtin/core-nodes/n8n-nodes-base.errortrigger/).
2. Add a Slack node, Send a message, into #automation-alerts, using the
   workflow name, the error message and the execution link the trigger hands
   you.
3. Point each production workflow at it. Open the workflow, then Options,
   Settings, Error workflow, pick yours, Save. One error workflow can serve
   every workflow you have.

Worth knowing:

- Manual test runs never fire the error workflow. Only automatic executions do.
- The error workflow does not need publishing. It runs as long as a workflow
  points at it.
- Each workflow needs the setting on its own. There is no instance-wide
  default, though a community template attaches it in bulk through n8n's API.
- If your n8n runs on a private host, the execution link only opens from inside
  your network.

## 5. Other platforms

Pipedream emails once per error per workflow per day by default. A workflow
subscribed to the Workspace `$error` events source receives every error as it
happens, with the original event attached, and a Slack step forwards it.

Activepieces sends project alerts by email only, a handful an hour at most.
Forward the email, or post from inside the flow when a step fails.

Power Automate emails a flow's owners about known, fixable failures and then
holds off for 28 days on that flow. For everything else, add a branch that runs
only when a step has failed and post through the Slack connector.

## 6. What Ray does with the channel

Ray reads the alerts as they arrive, like any channel it is invited to. It stays
quiet unless something clearly needs a person, and it asks before it changes
anything. Ask it what happened.

> **Priya:** @Ray what failed overnight in #automation-alerts? Group it by
> automation and tell me which ones are still failing.

Ray reads the night back, groups the repeats, and answers with which
automations failed, how many times, and the last error for each.

> **Marco:** @Ray file a Linear issue for the HubSpot lead sync failure, put the
> last error in the description, and assign it to Dana.

Creating the issue is a write, so Ray shows you the card first. The ticket
exists only after someone approves it.

> **Zach:** @Ray every weekday at 9:00, post a summary of overnight automation
> failures to #ops.

That is a scheduled digest. It runs without anyone asking.

## 7. What Ray will not do here

- It cannot re-run, pause, edit or switch off a Zap, a scenario or a workflow
  through this channel. Fixing the automation still happens in Zapier, Make or
  n8n.
- It sees only what the vendor posts. A Make run parked in Incomplete
  executions, or a Zap error that Zapier Manager missed, never reaches the
  channel and never reaches Ray.
- Reading a channel's history back is rate-limited by Slack, so a busy night
  takes Ray a few minutes to read back. Ask as alerts land and the answer is
  immediate.
- It does not page you. Ray may speak up when a message clearly needs a person,
  but for a push you can rely on, set the digest.
- It reads the alert text and nothing behind it. If the message says only
  "error", Ray can tell you an automation failed and not why. Put the error
  message in the alert.

## Questions people ask

**Do I need a paid plan on Zapier, Make or n8n?** Not for the alerts. A two-step
Zap runs on Zapier's free plan and the Zapier Manager step costs no tasks. Error
routes are part of Make's scenario editor. The Error Trigger is a core n8n node
in cloud and self-hosted.

**Does this use up my tasks or operations?** Each alert that posts costs what a
Slack step costs on that platform: one task on Zapier, one operation on Make,
one execution on n8n cloud. The trigger side is free on Zapier.

**Can Ray restart a failed Zap or scenario?** No. Ray can tell you it failed,
open the ticket and chase the owner. Restarting happens in the tool. If Ray gets
a direct connection to these platforms, this page will say so.

**Why did Ray miss one?** Almost always because nothing posted: Zapier Manager's
known issue, a Make run that parked as incomplete, an n8n workflow run by hand.
Check the vendor's own log first. If the alert is in the channel and Ray did not
see it, tell us through the support form at https://heyray.io/support.

**Does Ray need to be connected to Zapier, Make or n8n for this?** No. The whole
setup works with no integration at all, which is why it is the place to start.

**Can I use the channel Sentry or Datadog already post to?** Yes. Ray groups by
source when you ask. Keep conversation out of it.

That is the whole setup. One channel, Ray in it, and your automations reporting
to it. Add Ray to Slack at https://heyray.io/install and point the first alert
at it.