Check if your website implements the Web Model Context Protocol. We test for navigator.modelContext, .well-known/mcp.json, tool definitions, and W3C spec compliance. Instant results, no signup.
Enter any website URL. We’ll run six real-time checks against the W3C WebMCP draft specification and show you exactly where the site stands.
All checks run client-side in your browser. We don’t store your URL or any results.
Six checks mapped to the W3C draft spec published on February 10, 2026.
The discovery file that tells AI agents what tools your site offers. Like sitemap.xml for search engines, but for AI agent interactions. We check if this file exists and whether it contains valid tool definitions.
The core browser API that WebMCP relies on. We scan your page source for references to this API, which is how your site registers its tools with AI agents. Currently available in Chrome 146 Canary behind the “WebMCP for testing” flag.
Beyond the API itself, we check for supporting markup: meta tags that declare MCP capabilities, script includes that register tools, and any WebMCP-specific configuration in your HTML.
JSON-LD with Action types (SearchAction, OrderAction, ReserveAction) tells AI agents what your site can do even without full WebMCP. We check whether your structured data lays the groundwork for agent interactions.
Your robots.txt may be blocking AI crawlers like GPTBot, CCBot, or Google-Extended. While WebMCP operates at the browser level, blocking AI crawlers reduces your visibility in LLM-generated answers.
WebMCP requires a secure context. The navigator.modelContext API is only available on HTTPS origins, same as Service Workers, Web Crypto, and other modern browser APIs. No HTTPS, no WebMCP.
Gartner projects that 60% of web interactions will involve AI agents by 2028. Chrome 146 already supports WebMCP behind a testing flag, and Google developed the protocol alongside Microsoft and the W3C community group. The infrastructure is being built right now.
Today, when someone asks ChatGPT “find me a flight from Mumbai to Delhi on March 20,” the AI scrapes websites, reads text, and tries to piece together an answer from unstructured HTML. It works, sort of. It’s also slow, unreliable, and breaks when sites update their layouts.
WebMCP replaces that. Your site tells the AI agent: “Here are the functions you can call. Here are the parameters each one accepts. Here’s what you’ll get back.” The agent doesn’t need to parse HTML or guess at form fields. It calls your tools directly.
Sites with WebMCP get accurate, real-time responses. Sites without it get scraped, misread, or skipped entirely.
“WebMCP is the new robots.txt, but in reverse. Instead of telling bots what they can’t do, you’re telling AI agents exactly what they can do. The brands that define those tools first will own the agent-driven web.”
Hardik Shah, Founder of ScaleGrowth.Digital
If you’re an ecommerce brand, a SaaS product, a healthcare provider, or any business where customers take actions on your site, this matters. The checker above gives you a baseline. You’ll know where you stand before the standard moves from “testing flag” to “default on.”
We’ve published a full guide to WebMCP implementation on our WebMCP service page, including industry-specific tool architectures for ecommerce, healthcare, and SaaS.
Almost certainly not. WebMCP was published as a W3C draft on February 10, 2026, and is currently only available in Chrome 146 Canary behind a testing flag. As of March 2026, fewer than 200 websites globally have working implementations. Run the checker above to know for sure.
Not quite. Traditional APIs require developers to integrate with your documentation, authenticate, and write custom code. WebMCP uses the browser’s native navigator.modelContext interface. AI agents discover your tools automatically, the same way search engines discover your pages through sitemaps. You still need to design the tool architecture, but the discovery and connection layer is handled by the browser.
For a site with 3 to 5 core tools (product search, booking, pricing lookup), typical implementation runs 2 to 3 weeks. That includes the tool architecture design, JavaScript implementation, Chrome 146 testing, and security review. Larger sites with 10+ tools take 4 to 6 weeks. We cover the full process on our WebMCP implementation service page.
Because WebMCP adoption is still in the earliest stage. The spec was published in February 2026, and only Chrome Canary supports it behind a flag. Failing these checks puts you in the same category as 99%+ of the web. The point of this tool is to establish a baseline and show you exactly what to implement when you’re ready to move first.
We’ll design the tool architecture, build the JavaScript integration, and test it in Chrome 146. First movers get the advantage.