Free n8n error-handling template (workflow JSON) — stop silent failures

The most common way an automation dies is silently. Nobody notices for two weeks, then a client asks why an invoice never arrived. An error workflow is the cheapest insurance you can attach to any build.

Here's a small, importable n8n template I use as the baseline for client work — and the checks that matter more than the JSON itself.

Download

Download the workflow JSON (import into n8n).

It's four nodes:

Error Trigger → Format Failure Message (Code) → Post to Slack Webhook (HTTP, 3 retries)
                                             + sticky note with setup steps

The Code node reads the error-trigger payload defensively and produces a single message with the workflow name, last executed node, execution URL and the error message. The HTTP node posts it as {"text": "..."} — which works for a Slack incoming webhook, and can be repointed to Teams, Feishu, Telegram (different body shape) or your own endpoint.

How to use it

1. n8n → Workflows → Import from File → choose the JSON.
2. Open "Post to Slack Webhook" and replace the placeholder URL with your webhook.
3. Save the workflow.
4. For every workflow you want covered: Settings → Error Workflow → select this one.
5. Test it: temporarily break a target workflow (change a URL to something invalid), run it, confirm

the alert arrives with the execution link.

The part people forget: testing the failure path

An error workflow that has never fired is not an error workflow, it's a hope. The acceptance test for any automation should include:

• Deliberate failure — wrong token/URL/404 and check that the alert arrives within seconds.
• Execution link — the alert must contain a clickable execution URL, otherwise you can't debug.
• Retry policy — a transient 429 should retry before alerting; a 400 should alert immediately.
• No secret leakage — the alert must not include API keys or full payloads with personal data.
• A heartbeat — a workflow that stops running (deleted trigger, expired token) doesn't throw an

error. Add a daily "last successful run" check for anything mission-critical.

What this template deliberately does not do

• It doesn't retry the failed workflow itself (n8n has retry settings on the original workflow; blind

auto-retry can duplicate writes).

• It doesn't include credentials — you attach your own Slack/HTTP credential or use a webhook URL.
• It doesn't send emails by default; SMTP credentials vary by client. Add an Email Send node if needed.

Five checks before calling a workflow "done"

1. Error workflow bound and tested.
2. Idempotency: re-running doesn't duplicate records.
3. Timeouts on every external call.
4. Cost cap on any LLM node.
5. A runbook line that says who is woken up by the alert.

If you want the checklist I use for client handovers (acceptance tests, runbook, credential rotation), it's in the free audit call — or email me and I'll send the one-pager.

---

Workflow template by FlowSmith — fixed-scope automation, first prototype in 48 hours. If an n8n build has gone quietly unreliable, send me the workflow and I'll tell you what's missing.