Find out a client workflow broke before the client does.
InstanceDesk watches every n8n instance you look after, from outside. It emails you within minutes when a workflow fails or hangs, and keeps every earlier version of every workflow so you can see what changed and put it back. Alerts for workflows that quietly stop running are next.
Connects through n8n's public API. No agent to install. Credentials are never read.
| Instance | Status | Open incident | Failures, 24 h | Last check |
|---|---|---|---|---|
| Northside Dental | ||||
| Reminders | Needs a look | Stuck "Appointment reminders" has been running for 2 h 10 min | 0 | 40 s ago |
| Harbor Logistics | ||||
| Production | OK | None | 0 | 35 s ago |
| Staging | Failing | Failed "Lead intake" failed at node "Push to CRM", 3 times since 09:12 | 3 | 20 s ago |
| Brightleaf Studio | ||||
| Main | OK | None | 1, recovered | 50 s ago |
Know when it breaks, on every client instance
These parts are built and tested. Add an instance and every workflow on it is watched from that moment, with nothing to add to the workflows themselves.
An execution fails
One email per incident, not one per failure, naming the instance and workflow, and a second email when the workflow recovers.
An execution hangs
An alert when a run is still going after an hour, or whatever limit you set for that workflow.
Every change is kept
A backup each time a workflow changes, a readable diff between any two versions, and restore. More below.
Next: the failures that never throw an error
An error alert only fires when something runs and fails. The failures that lose clients are quieter. These are in build and will ship before public launch.
A workflow goes quiet
InstanceDesk will learn how often each workflow normally runs, weekday patterns included, and alert when the rhythm breaks.
Someone switched it off
An alert when a workflow is deactivated, archived or deleted, naming which one and when.
Slack and webhook alerts
Alerts go by email today. Slack and webhook delivery, routed per client, are coming.
See what changed. Put it back.
n8n keeps 24 hours of workflow history on most plans. InstanceDesk takes a backup every time a workflow changes and keeps it off-site, encrypted.
Pick any two versions and read the difference as a list of nodes, not a wall of JSON. Restore as a draft first, so nothing live changes until you have looked. A deleted workflow can be recreated from its last backup, and everything on an instance can be downloaded as a ZIP.
A restore brings back nodes, connections and settings. Tags and whether the workflow is switched on are not restored, because n8n does not allow that through its API.
Changes are picked up every 15 minutes, so several saves in that window become one version.
Lead intake · 3 Nov 14:10 → 9 Nov 09:42 · illustration
- ~ HTTP Request "Push to CRM": url changed
- ~ IF "Has phone number": condition changed
- + Slack "Notify sales" added
- − connection: Push to CRM → Send confirmation
- layout changes only: 3 nodes
A service you resell, not a cost you carry
On the Agency plan, monitoring costs you under $8 per client instance a month. Include it in a care plan and it pays for itself with the first client.
| Example | Per month |
|---|---|
| 10 clients on a $50 care plan that includes monitored upkeep | $500 in |
| InstanceDesk Agency, 10 instances | $79 out |
The $50 is an example figure, not a market rate. Use your own.
One API key. Nothing installed.
1. Create a key in n8n
Under Settings in your instance. If n8n offers scopes, read access to workflows and executions is enough for monitoring and backup.
2. Paste the address and key
InstanceDesk tests the connection and starts its first sync. It uses n8n's public API. Tested so far on self-hosted n8n 2.47.4; n8n Cloud is being tested before launch.
3. That is all
No node to add, no workflow to edit, nothing to keep updated on the client's instance.
We never read n8n credentials and never store the data your workflows process. The full list of what is called and stored is on one page: what we store and what we call.
Less than one billable hour a month
Free
- 1 instance
- Checked every 15 minutes
- Email alerts
- 7 days of version history
Solo
- 3 instances
- Checked every 5 minutes
- Email alerts
- 90 days of version history
Agency
- 10 instances
- Checked every minute
- Email alerts; Slack and webhook when they ship
- Unlimited version history
- Client groups
Studio
- 50 instances
- Checked every minute
- Email alerts; Slack and webhook when they ship
- Unlimited version history
- 90-day execution log
Planned launch plans and prices, in US dollars. Nothing is open for sign-up yet; the free plan opens after the private beta. Paid plans will start with a 14-day trial of Agency, without a card.
Join the waitlist
The first ten agencies get 30 days free in the private beta, in return for one feedback call. Everyone on the list hears before public launch.
You will get one personal reply, not a drip campaign.
Or pre-order: $79, refundable
Pays for your first month on the Agency plan and puts you at the front of the beta. Ask for it back at any point before your paid month starts and you get all of it. Your price stays $79 a month for as long as you stay subscribed.
Pre-order for $79What agencies ask first
I can build an error workflow in an hour. Why pay?
You should build it, and we give you ours as a free template. It covers failed executions and a simple "did it run?" check on one instance. It runs inside the instance it watches, it has to be installed and kept in step on every client instance, and it keeps no history of your workflows. InstanceDesk watches from outside, covers every instance from one place, and keeps every version.
Do you get access to my clients' credentials or data?
No. Credentials are never read, and the data passing through executions is never stored. Error messages are not captured unless you switch that on for an instance. A workflow backup does contain whatever was typed directly into its nodes, which is why backups are encrypted. The full list is here.
Which parts work today?
Built and tested: failure and stuck-run alerts by email with recovery notices, a backup whenever a workflow changes, the node-level diff, ZIP export, and restore as a draft, as live, or by recreating a deleted workflow. In build: alerts for silent, deactivated and deleted workflows, Slack and webhook delivery, and billing.
Which n8n does it work with?
It uses n8n's public API, which n8n offers on paid Cloud plans and on self-hosted editions including Community; n8n's free trial does not include it. So far InstanceDesk has been tested on self-hosted n8n 2.47.4 Community only. n8n Cloud is being tested before launch.
Will silence detection work for workflows with irregular traffic?
Failure and stuck-run alerts work for every workflow. Silence detection, when it ships, needs a rhythm to learn from: at least eight production runs in two weeks. For irregular workflows you will set the limit yourself or leave it off.
Is this made by n8n?
No. InstanceDesk is an independent product for n8n and is not affiliated with n8n GmbH.
Who is behind it?
One person who runs n8n instances for clients and needed this for them first. Support is by email, answered the next business day.