title: "Google ships WebMCP preview in Chrome, extends agent protocol to web" slug: "google-ships-webmcp-preview-in-chrome-extends-agent-protocol-to-web" published: "2026-07-16" beat: "Tools" tags: ["Tools", "Launches"] creator: "Agentry Newsroom" editor: "Susanne Sperling, Editor — Human in the Loop" tools: ["Claude (Anthropic)", "Perplexity Sonar"] creativeWorkStatus: "verified" dateReviewed: "2026-07-16" aiActArticle50: "compliant" humanView: "https://agentry.news/google-ships-webmcp-preview-in-chrome-extends-agent-protocol-to-web" agentView: "https://agentry.news/agent/google-ships-webmcp-preview-in-chrome-extends-agent-protocol-to-web"
Google and Microsoft unveiled an early preview of WebMCP, a W3C-standardized protocol for structured agent interaction with websites, shipping in Chrome 146 Canary in February 2026. The move extends t
Drafted by an AI agent. Verified by Susanne Sperling, Editor — Human in the Loop. AI policy.
Google shipped an early preview of WebMCP (Web Model Context Protocol) in Chrome 146 Canary during February 2026, marking the first browser implementation of a standardized protocol designed to let AI agents interact with websites through declared JavaScript tools rather than DOM automation W3C Web Machine Learning Community Group. The protocol was developed jointly by Google and Microsoft through the W3C, with specification editors including Brandon Walderman from Microsoft and Khushal Sagar and Dominic Farolino from Google Web Works.
WebMCP provides a JavaScript API, navigator.modelContext, that allows a web page to register tools available to an in-tab agent. Instead of an agent attempting to navigate the DOM—clicking buttons, reading text nodes, filling forms—it can invoke named functions directly, reducing brittleness and enabling more reliable agentic workflows. The draft specification was published on February 10, 2026 Agent Patterns as a Draft Community Group Report, establishing a formal foundation for browser vendors and developers to implement the standard.
The early preview in Chrome 146 Canary requires developers to enable the "WebMCP for testing" flag, limiting initial exposure to those actively testing the feature. Google plans a broader Origin Trial from Chrome 149 through Chrome 156, allowing a wider set of developers to experiment with WebMCP in production-like contexts without requiring users to flip flags. This staged rollout mirrors Google's standard approach to shipping experimental web APIs.
As of May 2026, adoption by mainstream conversational agents—Claude, ChatGPT, Gemini—remains at 0%, according to industry surveys Agent Patterns. This reflects the nascent state of the standard and the time required for both browser vendors and AI service providers to integrate WebMCP support into their platforms. The lack of immediate uptake also suggests that agent-website interaction patterns are still evolving; developers and model providers are evaluating whether WebMCP or alternative approaches (like structured APIs or LLM-native protocols) will become the dominant mechanism for agentic web commerce.
WebMCP complements existing agent infrastructure frameworks like the Model Context Protocol (MCP), which standardizes tool calling between agents and servers. By bringing structured, declarative tool registration to the browser, WebMCP addresses a gap: agents deployed in-tab or via cloud endpoints can now interact with websites in a more robust, maintainable way than screen scraping or DOM manipulation. This is particularly relevant to the emerging agentic commerce category, where agents need to reliably execute transactions, retrieve data, and manage workflows across multiple web properties.
The standard's development also signals that Google and Microsoft view agent-website interaction as important enough to standardize, even as the broader agentic AI ecosystem remains fragmented. Success will depend on developer adoption and integration by AI model providers—neither of which is guaranteed at this stage of the market.