The web is entering a new phase where AI agents are expected to do more than simply read information. Instead of browsing websites manually, AI agents could increasingly search, interact with forms, manage products, and complete tasks directly inside web applications. One technology gaining attention in this shift is WebMCP, or Web Model Context Protocol.
The latest WebMCP specification was published as a Draft Community Group Report on September 9, 2026, highlighting the continued development of a browser-based approach for connecting websites with AI agents. The specification is being developed through the Web Machine Learning Community Group, with contributors from major technology companies including Google and Microsoft.
What Makes WebMCP Different?
Traditional browser agents often have to understand a website visually or inspect its page structure before deciding which buttons and fields to use. This can make automated interactions slower and more fragile.
WebMCP takes a different approach. Websites can expose specific JavaScript-based tools that AI agents can discover and use. For example, an online store could expose tools for searching products, checking availability, or adding an item to a cart.
This gives an AI agent a more structured way to interact with a website instead of guessing how the user interface works.
Chrome and Cloudflare Push WebMCP Forward
WebMCP is currently being developed as an experimental browser technology. Chrome has been one of the major platforms supporting its development, while Cloudflare recently introduced a developer preview that allows websites to add a WebMCP interface without changing their origin application.
Cloudflare described WebMCP as a way for browser agents to work with websites through defined tools rather than relying entirely on traditional scraping and browser automation techniques.
The technology is also receiving attention from the broader developer ecosystem. Cloudflare's Browser Run supports WebMCP, allowing experimental browser sessions to discover and execute website tools directly.
A New Challenge for Website Developers
The emergence of WebMCP could change how developers think about websites. In the traditional web, developers primarily design experiences for human visitors. With agentic web technologies, websites may also need interfaces designed specifically for AI agents.
However, WebMCP is still evolving. Developers need to consider security, permissions, user confirmation, and potentially malicious instructions when allowing AI agents to interact with website functionality. Chrome's documentation specifically highlights the risk of indirect prompt injection and recommends careful security practices when exposing tools.
What Comes Next?
WebMCP is not yet a finalized web standard, but its continued development signals a broader shift toward an agent-ready web. If the technology matures and gains wider browser and developer support, websites may eventually become much easier for AI agents to operate.
For businesses, developers, and SEO professionals, this could represent an important change: the future website may need to be designed not only for people and search engines, but also for AI agents capable of taking action.
The next stage of the web may therefore be less about AI simply finding information—and more about AI actually getting things done.