Your AI builds it.
Vigilia keeps it running.
Vigilia watches your live app, explains every problem in plain language, and hands the fix straight to your AI — Lovable, Bolt, Cursor, Claude Code. Then it watches real traffic to confirm the fix actually held.
Checkout breaks when the cart is empty
Anyone who clicks “Pay now” with an empty cart sees a frozen screen instead of a message.
Who's affected: 14 users hit this in the last hour, mostly on mobile.
An incident in plain language, the fix for your AI — and proof it held.
From "it's broken" to "it's fixed" — and proof.
Install one script tag in under two minutes. After that, every problem runs the same four steps, and the last one is the step other tools leave to hope.
-
01
Vigilia spots it
Crashes, failed requests, downtime, slow pages — with what each visitor did right before.
-
02
Explains it
What broke, who it hits and how urgent it is, in plain language. Never a stack trace you have to read.
-
03
Your AI fixes it
A fix prompt written for your builder — or your AI fetches the incident itself and fixes the code.
-
04
Vigilia confirms it
It keeps watching real traffic, and tells you whether the fix held — or that it came back.
The step that proves it
Connect once. Then just ask.
Vigilia plugs into the AI you already build with. Ask it to fix your worst incident: it reads the diagnosis and the evidence, fixes the code, and hands it back to Vigilia to verify. No copying, no pasting, no context lost on the way. Built with Lovable today and Cursor tomorrow? Vigilia watches both the same way.
Built on MCP, the open standard for connecting AI tools. Prefer scripts? The same access is a REST API.
BookingSummary reads the total before it loads. I've handled the loading and empty states and added a test.Calm when all is well. Clear when it isn't.
Uptime, every minute on paid plans
Know your app is down before your users tell you. Give them a public status page and badge to check for themselves.
Alerts where you are
Email on every plan. Slack, Discord and signed webhooks on Pro, each with its own filters.
Quiet while you build
Errors from your laptop and preview links are recorded, never alerted — so the noise of building stays out of your inbox.
Open to your tools
An MCP server for your AI, a REST API for your scripts, and a Claude Code plugin with /vigilia:fix.
Every client app, watched — without watching them yourself.
You shipped it with AI; now it has to stay up. Vigilia tells you what broke before your client does, hands your AI the fix, and gives you proof to show them: a status page, a weekly report, fixes confirmed with real traffic.
A status page per client
Public, with their name and 90 days of uptime, plus a badge for their footer.
Alerts to the right channel
Each app routes to its own Slack, Discord or webhook.
A report worth forwarding
Health, uptime, what broke and what was fixed — every Monday, per app.
Your team, or your client
Invite them by email as members who see and fix, or admins who also change settings.
Private by default. Scrubbed before it leaves the page.
Emails, tokens and card-like strings are stripped out in your users' browsers — before anything is ever sent to Vigilia. Privacy isn't a setting here; it's how the snippet works.
Start free. Upgrade when it pays for itself.
- 1 app
- 10 AI diagnoses every month
- Plain-language error feed
- Uptime checks every 5 minutes
- Email alerts
- Public status page and badge
No card needed.
- 5 apps
- AI diagnosis on every error
- Copy-paste fix prompts for your builder
- Uptime every minute, plus 5 key pages per app
- Re-diagnose any incident on demand
- Slack, Discord and webhook alerts
- Server errors from your backend
€252 billed yearly. Cancel anytime.
Good questions, straight answers.
Do I need to know how to code?
No. Every problem is explained in plain language — what broke, who it affects, how urgent it is — with a fix prompt written for your builder. Paste it into Lovable, Bolt or Cursor, or connect your AI and let it fix things directly. The technical detail is there when you want it, never in your way.
Which builders and tools does it work with?
Any web app, wherever it's hosted. Fix prompts are tuned for Lovable, Bolt, v0, Cursor, Claude Code and Replit, and your AI can connect to Vigilia directly from Lovable, Bolt, Cursor, Claude Code, Windsurf, VS Code or Claude Desktop.
How do I install it?
Paste one script tag into your app's HTML — or tell your builder "add this script tag". Vigilia gives you a ready-made instruction for your AI, checks that the tag is live and allowed to run, and a one-click test link shows your first incident in seconds.
Can my AI fix incidents on its own?
Yes. Connect Vigilia to the AI you build with (it speaks MCP), then ask it to "fix my worst Vigilia incident". It reads the diagnosis and the evidence, fixes the code, and hands it back — and Vigilia then checks with real traffic whether the fix held.
What counts as an AI diagnosis?
Only a problem that needs a model to explain it. Errors Vigilia already knows how to explain are always free, and each distinct error is diagnosed once, not every time it happens. Free includes 10 AI diagnoses a month; Pro diagnoses every new error.
Will it slow my app down?
No. The script is under 10KB, loads after your page, batches what it sends, and is built so it can never throw an error into your app.
Is my users' data safe?
Emails, tokens and card-like numbers are stripped in your visitors' browsers before anything is sent, and stripped again on arrival. Everything is stored in the EU. You can export or delete all your data from Settings at any time.
What happens after I mark something fixed?
Vigilia keeps watching. If the error stays away through 24 hours of real traffic, you get a "your fix worked" message. If it comes back, the incident reopens and tells you the fix didn't hold.
Your app is live. Know when it breaks — and when it's fixed.
Free for your first app. One script tag, under two minutes.