WebMCP (Web Model Context Protocol) is Google and Microsoft's new browser-based standard launched in Chrome 146 Canary on February 10, 2026. It allows your website to publish a structured "Tool Contract" via a new JavaScript API (navigator.modelContext), telling AI agents exactly what actions they can perform — like search, book, or buy — instead of guessing from screenshots or DOM elements. It matters because AI agents acting on behalf of users are rapidly replacing traditional browsing. Websites that aren't agent-ready will lose traffic and conversions to those that are.

Use the free WebMCP Checker tool on this page. Enter your URL and click "Run WebMCP Audit." Our AI analyzes your site across 8 key dimensions: HTML form quality, semantic markup, JavaScript API readiness, Tool Contract potential, form action clarity, structured data, authentication compatibility, and agent-discoverable actions. You'll get a score (0–100), specific fix recommendations, and ready-to-use implementation code tailored to your website type.

Anthropic's MCP (Model Context Protocol) is a server-side protocol using JSON-RPC, designed for backend service-to-service automation where no browser UI is needed. WebMCP runs entirely client-side inside the browser tab and is designed for user-present interactions. Think of it this way: MCP is for when no human is watching; WebMCP is for when the user is present. They are complementary — a travel site might run a backend MCP server for direct ChatGPT/Claude integrations while also implementing WebMCP on its consumer-facing booking flow.

WebMCP is currently available in Chrome 146 Canary (February 2026) behind the "WebMCP for testing" flag. Microsoft Edge support is expected soon given co-authorship of the spec. Safari (WebKit) and Firefox (Gecko) have indicated support timelines of late 2026 and early 2027 respectively. Full stable Chrome/Edge rollout is anticipated mid-to-late 2026 with Google I/O and Google Cloud Next as likely announcement venues. The W3C formal draft transition is underway. Use the @mcp-b/global polyfill for cross-browser compatibility today.

Two methods: (1) Declarative API — add toolname and tooldescription attributes to existing HTML forms. Minimal effort if forms are well-structured. (2) Imperative API — use navigator.modelContext.registerTool() with a name, natural language description, JSON schema for inputs, and an execute callback function. Also add a /.well-known/webmcp JSON manifest for autonomous agent discovery. Use this tool's Code tab to get tailored implementation code for your site type.

Absolutely. SEO experts have called WebMCP "the biggest shift in technical SEO since structured data." As AI agents increasingly act on behalf of users, a new discipline called AEO (Agentic Engine Optimization) is emerging. Where SEO optimized for "how does my content rank in search?", AEO optimizes for "can AI agents reliably take actions on my site?" Sites with WebMCP earn higher agent invocation rates, better execution success rates, and more AI-driven conversions. Early adopters gain significant competitive advantage.

Score guide: 75–100 (Agent-Ready) — strong foundations, begin WebMCP implementation now. 50–74 (Good Foundation) — solid base, key gaps to address before implementation. 25–49 (Needs Work) — significant structural improvements needed first. 0–24 (Not Agent-Ready) — fundamental HTML/JS issues to resolve before WebMCP is viable. Use this tool's Fix It tab to get prioritized recommendations for your exact score band.

Not yet in production browsers. Currently WebMCP is only available in Chrome 146 Canary behind a feature flag. However, you can implement WebMCP today using the polyfill (npm install @mcp-b/global) which provides cross-browser compatibility. Your regular site visitors won't notice any difference — WebMCP tools are invisible to users and only activated when an AI agent invokes them.