In the agentic web, a broken site fails silently. That's why we built SiteVitals.
SiteVitals is a website health monitoring platform. If you're an agency or developer responsible for multiple websites, it gives you a single place to see health data across your entire portfolio: uptime, SEO, security, performance, SSL, domain expiry, asset integrity, content changes, all plotted on a unified timeline.
That's it in a nutshell, here's why it's more relevant now than ever.
The background
We've been building and supporting websites for over two decades, with clients ranging from sole traders to the NHS, BBC Two, and Aardman. In that time we've developed a fairly comprehensive picture of what websites need to stay healthy, and what it looks like when something isn't.
Uptime monitoring is the standard starting point, but it only tells you whether your server answered. A 200 response doesn't tell you whether your canonical tags are intact, whether your PageSpeed score has been sliding, whether a security header was dropped in a recent deployment, or whether your SSL cert is 12 days from expiry. There's a lot of ground between "the server is up" and "the site is healthy", and most monitoring tools don't cover it.
SiteVitals covers that ground continuously, across as many sites as you're managing.
Why it matters more now
In the old web, a broken site was visible. A missing image, a mangled layout, a form that didn't submit. A real human would see it and complain. That complaint was your signal. Annoying, but at least you knew.
The agentic web changes that feedback loop entirely. AI agents don't complain. Broken structured data gets ignored. A slow API response causes the agent to move on. Missing schema markup means your content doesn't get surfaced. None of this throws an error you'll see. There's no angry email from a visitor. Your site keeps returning 200s, your uptime monitor stays green, and your leads dry up while you're looking the other way.
This is why external, independent monitoring matters more now than it ever did before. You need something watching the things that agents rely on, not just the things humans notice. And you need it watching continuously, because the signal that something is wrong is no longer going to come from your users.
Beyond that, the surface area of things that can change on a website is growing. Marketing teams are shipping landing pages without going through a developer. Agents are touching production systems. The amount of code going live has gone up considerably, and the human attention available to review it before it ships has not. You don't need to be inside the repo to check whether a site is fast, secure, technically sound, and findable. That's the job SiteVitals does. The technical backbone performing constant health checks on your web portfolio, independent of whoever or whatever shipped the last change.
What we built
Nine monitoring areas running continuously:
Uptime - checks every minute, free tier available
SEO - broken links, meta issues, canonical tags, AI crawlability
Security - headers, mixed content, vulnerabilities, form injection
Performance - PageSpeed scores tracked over time, not point-in-time snapshots
SSL and domain expiry - flagged well in advance, not after the fact
Asset integrity - third-party scripts, fonts, external dependencies
Content monitoring - unexpected changes on key pages
Cron job monitoring - silent failures in scheduled tasks
Server monitoring - CPU, RAM, disk, and load tracked continuously via a lightweight agent, with threshold alerts and service monitoring
Everything lands on a single unified timeline, including your own deployment notes and, if you're running WordPress, core, plugin and theme updates pushed automatically via our WordPress plugin. When something shifts in your scores, you can scroll back and see exactly what was happening at the time. Context rather than guesswork.
Everything comes together on the timeline. A PageSpeed score of 75 means very little on its own. A PageSpeed score that moved from 91 to 75 over six weeks, with a notable drop the day after a plugin update, tells you exactly what to investigate and when the problem started. That historical context is what turns monitoring data into something actionable.
In the last 90 days alone, SiteVitals has run over 5.1 million checks across 449 active monitored sites, logging 14,904 site events on customer timelines.
For agencies, there's another angle worth mentioning. The data doesn't just tell you what needs attention on a client site, it gives you a concrete metric to start a conversation with. A PageSpeed score that's dropped, a security issue that needs addressing, a domain expiring in 30 days are the kinds of findings that open up conversations about billable work naturally, without anyone having to go looking for it.
MCP: letting AI assistants query your site health directly
We've built a REST API and an MCP (Model Context Protocol) server, so AI assistants like Claude can query your live SiteVitals data directly. You can ask "which of my sites has the worst SEO score?" or "is SSL about to expire on any of my domains?" and get an answer pulled from real, current data.
The obvious question is: why not just let the LLM run the checks itself? Two reasons.
The first is history. An LLM running a check right now has no idea what your PageSpeed score was three weeks ago, or that it dropped the day after a deployment. The timeline is the product, and you only have it if something has been watching continuously.
The second is cost. Running constant health checks across hundreds of sites using tokens would be prohibitively expensive. SiteVitals does that work with a rules-based engine, stores the results, and the LLM queries the output in a single cheap call. The LLM gets the intelligence layer; we do the continuous grunt work. It's a sensible division of labour and it's where we think website portfolio management is heading.
We also launched a Firefox extension last week, approved by Mozilla, that lets you run an AI-powered audit of any site directly from the browser toolbar.
How we built it
We're a small team with no outside funding. We use LLMs heavily in the build process. Whilst they need help with architecture, the power of their code generating abilities has meant we've been able to deliver a product that's far exceeded our initial expectations. The stack is Laravel and Livewire. In a world of overcomplicated JS frameworks, we chose a stack that lets us move fast and stay reliable. Lisa handles the product and feedback loop; I handle the technical architecture.
Where we are
Launched: late 2025
Active sites monitored: 449
Checks run in last 90 days: 5,179,559
Site events logged in last 90 days: 14,904
Pricing: £19/month, or £15.17/month paid annually. We lead with the annual figure because it's the better deal and we'd rather people see that upfront.
A few things I'd love to hear from this community
We'd love to get feedback on what we've built. For agencies managing a large number of sites I think it's a powerful dashboard. I'm particularly excited about how LLMs can use the timeline data to help agencies manage, improve and advise on site health, turning raw monitoring data into genuine actionable intelligence.
How many of you find out a client's site is broken from the client first, even with uptime monitors running? That's the gap we're trying to close.
And if anyone is building in the MCP space, whether as a data provider like us or on the agent side, I'd be interested in where you think the practical limits are right now. We're treating it as a first iteration and there's a lot of ground still to cover.
https://www.sitevitals.co.uk MCP docs: https://www.sitevitals.co.uk/docs/mcp
