Google Chrome ships WebMCP, turning every website into a tool for AI agents (venturebeat.com)

Google Chrome has introduced WebMCP, the Web Model Context Protocol, in an early preview with Chrome 146 Canary. This proposed web standard, co-developed by Google and Microsoft through the W3C's Web Machine Learning community group, allows websites to expose structured tools directly to AI agents via the navigator.modelContext API. This advancement signifies a substantial improvement for the AI/ML community, enabling AI agents to interact more effectively with web applications without the inefficiencies of traditional web scraping or DOM parsing, which often lead to high costs and latency. WebMCP offers two main APIs: the Declarative API for handling standard HTML form actions and the Imperative API for more complex interactions, enabling developers to turn existing JavaScript logic into agent-readable tools. As a result, tasks that usually require multiple browser interactions can be distilled into a single structured function call, drastically reducing token consumption, improving reliability, and speeding up development by leveraging existing client-side code. Designed for collaborative workflows rather than full autonomy, WebMCP emphasizes a human-in-the-loop approach, enhancing user-agent interactions while aiming for broader adoption and standardization in the web and AI landscape.