Make your homepage
clearer to AI agents.

Find business details we can read on your public site, such as services, prices or delivery information. Get suggested fixes for signals we do not find; the scan does not test bookings, purchases or payments.

Agent Discovery API — example response

# Query the live registry from any agent framework

$ curl "https://mcphubz.com/api/discover?category=database&minScore=80"

{ "meta": { "total": "<matches>", "count": 2 },

"servers": [

{ "name": "...", "category": "database", "qualityScore": 0–100 },

{ "name": "...", "category": "database", "qualityScore": 0–100 }

] }

✓ Live endpoint · CORS enabled · JSON and MCP-native format

# Names and scores vary by query; the numbers above are placeholders.

Free
Site Scan
no signup required
0–100
Readiness Score
scored against your industry
Weekly
Re-checked
certification stays honest
1,100+
MCP Servers
as a supporting resource

A score is not a certificate.

The scan helps you improve what your homepage explains. Certification adds verified site ownership, the MCPHubz snippet and passing required checks. Neither is a guarantee that an AI assistant will recommend your business or complete a transaction.

What our scan tells you

Which business-specific signals on your public pages we found, and which we did not. For example: services for a plumber, menu information for a pizza shop, or product details for a store. A missing signal means we did not find enough evidence, not that your business lacks it.

What certification records

Verified site ownership, the MCPHubz snippet and passing required homepage checks at the last readable assessment. It is not a test of booking, checkout or payment. Weekly checks are scheduled; they are not continuous monitoring. Use the public verification page to see the recorded status and check date.

Breaking: WebMCP

WebMCP: What the
Draft Actually Proposes

WebMCP is the browser-native evolution of the Model Context Protocol. Instead of an agent screenshotting your page and guessing at buttons, your site exposes named tools it can call directly. No server, no scraping. It is an early draft, available through a per-site origin trial in Chrome 149, and we are tracking every revision.

  • Browser-native AI agent protocol — no server required
  • W3C Community Group Draft Report, co-authored by Google and Microsoft engineers
  • Chrome 149 origin trial - sign up per site
  • Two APIs: Declarative HTML + Imperative JavaScript
  • Aims to replace screenshot-based browsing with direct tool calls
  • A site that registers tools becomes callable by an agent that supports it
Read the Full WebMCP Guide
webmcp-example.js

// Register a tool on your webpage

navigator.modelContext

.registerTool({

name: "search_servers",

description: "Find MCP servers",

execute: async (params) => {

return await searchDirectory(params);

}

})

// Illustration for your own page. The API it calls is live today.

Certification is free

No card or trial. Start with the free scan, then verify your site and address any required gaps before certification.

$0free while we are early

Free scan and conditional certification. Website changes are not made automatically.

  • Business-specific homepage findings and suggested fixes
  • Required checks must pass before readiness certification
  • Certificate of Authenticity and badge when eligible
  • A public verification link showing recorded status
  • Scheduled weekly checks, not continuous monitoring

We may be blocked by a site or receive only part of its homepage. We explain those limits rather than treat unreadable content as a passing scan.

See what your homepage
needs to explain.

Start free with business-specific homepage findings and suggested fixes. Blocked or partial reads limit the assessment; no bookings, purchases or payments are tested.