Reliability monitoring for agencies running n8n. Grade any workflow free below — then let us watch your client workflows and alert you before they notice.
Export a workflow from n8n (⋯ menu → Download) and paste it, or drop the file in. Seven scored checks, each mapped to a documented way a run finishes green having done nothing — plus one advisory the export can't tell us. You get a score out of 100 and the specific fix for every check you fail. Nothing is uploaded; the analysis runs in your browser and works with your network off.
n8proof rewrites your workflow JSON with retries enabled, error outputs wired to a real handler, and guards on unvalidated webhooks — then hands you the file back. It also builds the error workflow your export doesn't have, ready to import.
Review the diff before you import. This is a static rewrite of your JSON — it cannot know your business logic. Import it as a new workflow, compare, and test before you replace anything a client depends on. Never blind-import a file a website gave you.
Five founding agencies, onboarded by hand — I wire the monitoring into your workflows with you and build the first client report alongside you. It's early and it's deliberately small. If you'd rather just use the free grader forever, that's a completely fine outcome and it stays free.
Where should I send it? These files are free. Leave an email and I'll send the file plus a short note on the two failure modes this repair still can't catch. One question I'd love answered: how do you find out today when a client workflow silently stops?
Your workflow is never uploaded either way. The email is so I can send the file and follow up once — not a newsletter.
Something noticed. The dangerous class is the run that stops doing its job with no error anywhere. So we measured how common that is: we ran the grader on this page across 2,066 public n8n workflow templates. These are the results.
Measured 31 Jul 2026 across 2,066 workflows from the public template corpus, using the exact grader above — same 7 checks, same weights. One correction applied before publishing: an eighth check (whether an error workflow is attached) was excluded from scoring, because no export in the corpus carries that field, so counting it would have manufactured a fake catastrophe. Templates are starting points, not production deployments — which is the point: agencies import these and ship them. Happy to send the full method and the reproducible harness to anyone who asks.
We watch executions for the failures n8n calls success — swallowed errors, empty outputs, counts that quietly collapse from 500 to 5 — and page you the moment one lands.
A weekly report per client, in your branding: workflows watched, runs verified, issues caught before impact. The document that makes a retainer renew itself.
The scariest failure is the run that never happened. If an expected run doesn't appear, you hear about it in minutes — not when the client does.
Silent failures are the most-cited unsolved problem in the n8n community right now — here's what agencies and builders are saying, unprompted.
"Runs n8n marks green but broke… the usual signal never fires. A run can quietly stop doing its job and nothing pages you."
"The failure most teams miss is silence. Agencies lose trust when a workflow quietly stops delivering — not when it throws an error."
"How do you monitor n8n workflows in production and detect failures early?" — 24 replies, no consensus answer.
Founding price is locked for life. Public pricing rises to $99/mo once the five founding spots fill.
Yes — and it's good. The free grader here is the same category, but n8proof's product isn't the grader. It's the ongoing monitor and the white-label client report, aimed squarely at agencies who need to prove reliability to the people paying the retainer. The grade is the front door; the report is the reason you stay.
An error workflow catches thrown errors. It does nothing for the failure class that matters most: runs that finish green having done nothing — swallowed errors, empty outputs, schedules that silently stopped. That's what n8proof watches, plus a dead-man's switch for the run that never fired at all.
No. Every check runs in your browser via JavaScript. The JSON you paste is never sent, stored, or logged anywhere. You can disconnect your network and it still works.
$49 for a hands-on 7-day setup sprint, with a satisfaction guarantee. Inside one week I wire monitoring into your client workflows with you, and we build the first white-label report together — on a call, not via a docs page. That also reserves one of five founding spots and locks your $49/mo for life (public price will be $99/mo). If the sprint doesn't leave you better off than when you started, you get the $49 back and you keep whatever we built. Being early here means shaping what gets built — the monitor is being built with these five agencies, not sold to them.
Five founding agencies, onboarded by hand. Your workflows wired in, reports in your branding, founding price locked at $49/mo for life. Starts with $49 for a hands-on 7-day setup sprint, with a satisfaction guarantee.