Bring your agent: Negative Zero is WebMCP-native from day zero
Any AI agent that can drive a browser can now get a Cyber Essentials Plus price, check your readiness, book a demo or ask about our security posture, via WebMCP, not by scraping our forms. And agents get 10% off, through 31 March 2027.
Most websites are built for humans and are merely tolerated by software. Ours is built for both, and it has been since the day zero of go-live.
Negative Zero is WebMCP-native. WebMCP is the emerging W3C standard, implemented by Google and Microsoft in Chrome and Edge, that lets a web page declare the things it can do as typed, callable tools, so an AI agent working on your behalf can discover and use them directly, instead of guessing which form field is which. On every page of negativezero.com, an agent with WebMCP support finds twenty tools registered through document.modelContext. Every one of them is documented in plain text at /llms.txt for agents that read rather than call.
What your agent can do here
Find things out:
Get things done:
Every action tool is marked consequential, tells the agent to confirm the details with you first, and is protected by the same bot check, rate limits and validation as our forms. The agent acts through the page you have open, not around it.
Agents get 10% off Cyber Essentials Plus, through 31 March 2027
To say thank you to the people bringing agents to our site first, quotes requested through the agent tool carry a 10% agent launch offer, running until 31 March 2027. It's applied automatically and shown on the quote against the list price.
Why we build like this
We are an agent-native cyber security company: agents already do work that used to take a team, and we build for that rather than against it. Our job is to use threat-informed prevention to implement fixes, patches and controls fast, and as a Certification Body, to make getting certified as painless as it should be. The Negative Zero Agent is built the same way, reading and acting through typed tools and real permissions, not by taking control of a screen and guessing where the buttons are. We think that is where every agent is heading, and we would rather build for it now than retrofit it later.
Most "AI browsing" today is a computer-use agent: it opens your site, reads a DOM tree that was never written for software, and clicks whatever looks right. That is a security problem because an agent reading raw page content has no reliable way to tell your instructions from something an attacker hid inside the page itself. WebMCP is the alternative: a site declares exactly what it can do, an agent calls the one tool it means to call, and there is no page left for an attacker to hide an instruction inside.
A site that fights agents helps nobody. A site that gives them precise, safe, clear tools, and never pretends a response has been released when a person still has to approve it, is how we want to be treated, and it is how we are building the Negative Zero Agent itself.
If you are building or using agents, try it: open negativezero.com in Chrome, or read /llms.txt. If you are seeking Cyber Essentials certification, ask your agent for a price from us!
