WebMCP is an emerging web standard, a new navigator.modelContext browser API, that lets a website expose its own actions to AI agents as structured, callable tools instead of the agent screenshotting the page and guessing where to click. It was authored by engineers at Google and Microsoft through the W3C Web Machine Learning Community Group, announced in February 2026, and entered a public origin trial in Chrome 149 in May 2026.

Is WebMCP the same as llms.txt?

No. llms.txt is a read interface: a static file that helps AI models understand and cite your content. WebMCP is an action interface: your site publishes callable tools an agent can invoke to actually do things, like get a quote, search, or book. Google's John Mueller has called llms.txt purely speculative and said he prefers the WebMCP approach because it has clear goals and processes.

Is WebMCP ready to use in production?

Not yet as a ratified standard. As of mid-2026 WebMCP is in a public origin trial in Chrome and lives in a W3C community group, not the official standards track. That is exactly why being early matters: you can build and test agent-ready tools now, before the standard ships broadly and your competitors move.

How do I make my website agent-ready with WebMCP?

You expose tools two ways: an imperative JavaScript API, navigator.modelContext.registerTool(), for dynamic actions like search or filter, and a declarative HTML annotation, data-mcp-tool, on forms you already have. The work is deciding which high-value actions to expose and describing them the way an agent expects. Digilu builds this for you.

WebMCP Agent-Ready Check

Demo · Beta

Enter your site and we fetch it live, looking for the two WebMCP surfaces. You get an honest verdict in seconds. Almost no site has this yet, so a clean “not yet” is expected.

AIOInsights Plus

WebMCP Live Demo

Is your site agent-ready?

A new browser standard, WebMCP, flips how sites talk to AI. Instead of an agent reverse-engineering your pages, screenshotting the screen and guessing where to click, your site hands it a menu of tools it can call directly. This demo checks whether your site exposes them yet.

Run the demo What is WebMCP?

Today: the agent guesses

An AI agent lands on your page and has to figure it out like a person would: read the screen, hunt for the right button, and hope the click does what it expected. Slow, brittle, and often wrong.

With WebMCP: you hand it the menu

Your site declares its actions as callable tools: get a quote, check availability, start a booking. The agent calls the tool directly, with the inputs it needs. No guessing, no screenshotting.

What WebMCP actually is

WebMCP is a new browser API, navigator.modelContext, that lets a website expose its own features to AI agents as structured, callable tools. It was authored by engineers at Google and Microsoft through the W3C Web Machine Learning Community Group, announced in February 2026, and went to a public origin trial in Chrome 149 in May 2026. It is emerging as the real business-to-agent interface for the agentic web.

Why this is not just another llms.txt

They solve different problems. One helps AI read you. The other lets an agent act on your site. That difference is why WebMCP has momentum where llms.txt stalled.

	llms.txt	WebMCP
What it is	A static text file at your root	A live browser API your site calls
What it does	Hints to AI what your content means	Publishes actions an agent can invoke
Read or act	Read only	Act: quote, search, book, submit
Adoption	Years old, largely unused by AI systems	Backed by Google + Microsoft, in Chrome origin trial

“I like the WebMCP approach, as well as the commerce integrations: they have clear goals and processes.”

John Mueller, Google, who separately called llms.txt “purely speculative for now.”

Honest status: WebMCP is in a public origin trial, not a finished standard. It works in Chrome behind a trial flag today and is not yet supported across all browsers. We are telling you this because being early is the point: you can be agent-ready before the standard ships broadly and your competitors notice.

Want to actually be agent-ready?

This demo tells you if you are agent-ready. Digilu builds the part that makes you agent-ready: the callable tools on your site, describing your highest-value actions the way agents expect. It is a managed add-on, and while WebMCP is in trial, we get you ready for the day it ships.

Talk to Digilu about WebMCP

Beta 0.1 · Found a problem or think we got something wrong? Help us make this better.