n8n Silent-Success Auditor

Most n8n monitoring watches for failures. An execution errors, the error trigger fires, somebody goes and fixes it. The run nobody catches is the other one: the schedule fires on time, every node goes green, and zero rows come out the far end. Nothing in n8n treats that as a problem, so the executions list fills up with ticks while the data quietly goes stale.

I built this after one of my own scheduled jobs spent weeks reporting success and writing nothing. This Actor reads the item counts inside your own executions and treats an empty successful run as a fault in its own right.

What it checks

For every workflow it scans, the Actor asks your n8n API for the last successful execution, opens that execution's run data, and counts the items the final node actually returned. Each workflow lands in one of five states:

State	Meaning
HEALTHY	ran inside its expected window and the final node returned at least the minimum items
EMPTY	the last run finished green and produced nothing
STALE	nothing has succeeded inside the expected window
DARK	no success inside three expected windows, or no successful run retained at all
UNREADABLE	the n8n API could not be read for this workflow, so its state is unknown rather than fine

Every row carries the workflow name, the state, why, minutes since the last success, the final node's name, its item count, the execution id and a link straight to that execution in your instance.

What you need

1. Your n8n instance URL, for example https://n8n.example.com.
2. An n8n API key. Create one in n8n under Settings, then n8n API. It only needs read access to workflows and executions. Apify stores it as a secret and it never appears in the log or the dataset.

The Actor works with self-hosted n8n and n8n Cloud. It uses the official n8n public API, so executions must be retained on your instance for it to read them (check your execution data settings if everything comes back DARK).

How to run it

Run it once with no API key and you get a demo dataset: one sample row per state, nothing charged, so you can see the output shape before connecting anything.

Leave the watch list empty and it scans every active workflow using the defaults: expected once a day, at least one item from the final node, and a 1.5 grace multiplier before a late run counts as STALE.

For finer control, give it a watch list:

[
  { "workflowId": "abc123", "label": "Nightly CRM sync", "expectEveryMinutes": 1440, "minItems": 1 },
  { "workflowId": "def456", "label": "Hourly price scrape", "expectEveryMinutes": 60, "minItems": 20 },
  { "workflowId": "ghi789", "label": "Weekly cleanup", "expectEveryMinutes": 10080, "minItems": 0 }
]

A workflow that legitimately returns nothing most of the time gets minItems: 0 and is judged on freshness alone.

Schedule the Actor on Apify to run every hour and wire its dataset or the OUTPUT summary into Slack, email or a webhook with an Apify integration. The OUTPUT record holds a one-line text summary, the tally per state and the list of faults with links.

Pricing

Pay per event: one charge per workflow scanned, nothing else. Scanning 20 workflows every hour costs 480 events a day. There are no proxies and no browser behind this Actor, so the platform usage on top is tiny.

Limits

• The Actor reads only the last successful execution per workflow. It does not read failed runs, because your error trigger already covers those.
• Item counts come from the final node executed. A workflow whose useful output happens mid-chain (for example a Slack post followed by a cleanup step) should set minItems to match what that final node returns.
• n8n's public API paginates executions and workflows; large instances take a few seconds per hundred workflows.

Prefer to self-host it?

The same logic ships as a free n8n template on my creator profile, https://n8n.io/creators/fractional-hq, with the verdict engine and the Slack alerts inside a workflow you run on your own instance. This Actor is the hosted version for people who would rather not babysit a second workflow.

Built by Fractional HQ

I run Fractional HQ, a one-man studio for design and automation. I do not write code by hand: I architect, review and hold the standard while AI agents do the typing. That method runs my own estate, seven brands, a Postgres operations system, more than 100 scheduled jobs, published n8n templates, Zapier Silver certified.

Need this wired into your own alerting, or want the whole monitoring layer built and handed over working? Get in touch: https://fractionalhq.uk/brief