What is WebMCP?
WebMCP is a proposed web standard that would let a website expose its own functions as structured "tools" — with natural-language descriptions — that an AI agent browsing the page could call directly, instead of the agent having to guess its way through clicking buttons and filling forms. It's still a draft, not a finished or widely supported standard.
ShoaibUpdated 3 min read
Who is behind it, and what stage is it at?
WebMCP is being developed as a draft specification by the Web Machine Learning Community Group, with editors from Google and Microsoft involved. The specification itself states plainly that it "is not a W3C Standard nor is it on the W3C Standards Track" — meaning it's an early proposal, not something ratified across the web.
What problem is it trying to solve?
Right now, an AI agent that operates a browser has to interpret a page the way a human would — reading the visual layout, guessing which button does what, and clicking through it step by step. WebMCP's idea is to let a site instead declare its own actions directly (for example, "add this to cart" or "check availability") as callable tools with clear descriptions, so an agent can act more reliably and directly.
Does any browser actually support it yet?
Support is early and experimental at best. There's no evidence of it being a finished, cross-browser standard — Firefox and Safari's teams are engaged in the spec discussion but haven't committed to shipping it. Treat WebMCP today as an emerging concept under active discussion, not infrastructure you can rely on.
Should a business in Kuwait care about this now?
Not urgently. It's worth being aware of as a direction the web may move in — agents interacting with sites more directly — but there's nothing to implement yet, since the spec itself is incomplete and not adopted broadly.
Want a website built to stay ahead of how search and AI agents actually work? See our Landing Pages service.