mintlify.com

5 tool(s) registered on document.modelContext.

• Tools: get_site_overview, navigate_page, read_page, read_skill, search_docs
• First tool registered 746ms after the page started loading.

No declarative forms, but tools are registered with JavaScript. You only need one of the two.

• Declarative forms are optional when tools are registered with JavaScript.

We called 4 tool(s) with test inputs based on their inputSchema. 1 returned an error: • search_docs: returned an error: "Search failed (500)"

An agent that calls a broken tool gets an error instead of an answer. This check fails if any tested tool returns an error.

• Working: get_site_overview
• Found nothing for our test inputs (not counted against you): read_page, read_skill

async execute({ query }) {
  try {
    const results = await search(query);
    return { content: [{ type: "text", text: format(results) }] };
  } catch (err) {
    // Mark the failure with isError. If the error message is returned as a
    // normal result, the agent reads it as the answer.
    return {
      isError: true,
      content: [{ type: "text", text: `Search unavailable: ${err.message}` }]
    };
  }
}

First tool registered 746ms after the page started loading.

• Within the 5 second limit.

All 5 tool(s) have a valid inputSchema with a description on every property.

• get_site_overview: no-argument tool, schema valid.
• navigate_page: 1 described property.
• read_page: 1 described property.
• read_skill: 1 described property.
• search_docs: 1 described property.

5 of 5 description(s) need work: • get_site_overview: description doesn't say what the tool returns: "Get an outline of the Mintlify documentation, grouped by section, with a short description of each page. May be truncate… (180 chars total)" • navigate_page: description doesn't say what the tool returns: "Navigate the current browser tab to a Mintlify documentation page." • read_page: description doesn't say what the tool returns: "Read the full content of a Mintlify documentation page as markdown." • read_skill: description doesn't say what the tool returns: "Read a Mintlify agent skill document by name." • search_docs: description doesn't say what the tool returns: "Search the Mintlify documentation for relevant pages and content."

Agents choose a tool based on its description. If the description doesn't say what comes back, the agent has to call the tool to find out.

- description: "Product search"
+ description: "Search the product catalogue and return matching product names, prices and stock status."

1 of 5 tool name(s) don't follow the convention: • navigate_page: starts with "navigate", which is not a verb.

Agents pick tools by what they do. A name that starts with a verb, like search_products, makes that clear. A noun like products doesn't.

• Following the convention: get_site_overview, read_page, read_skill, search_docs

- name: "ProductSearch"
- name: "products"
+ name: "search_products"    // snake_case
+ name: "searchProducts"     // camelCase — either is fine, be consistent

All 5 tool(s) have the annotations they need.

• get_site_overview: readOnlyHint=true, untrustedContentHint=true
• navigate_page: readOnlyHint=false, untrustedContentHint=false
• read_page: readOnlyHint=true, untrustedContentHint=true
• read_skill: readOnlyHint=true, untrustedContentHint=true
• search_docs: readOnlyHint=true, untrustedContentHint=true

Every tested tool returned between 1 and 20,000 characters.

• get_site_overview: 12,000 chars.
• read_page: 17 chars.
• read_skill: 21 chars.
• search_docs: 19 chars.

/llms.txt responded 200 with 6,479 bytes.

• /llms.txt → 200 text/plain; charset=utf-8
• First bytes: # Mintlify > Mintlify is the knowledge platform built for agents: self-updating documentation for startups, enterprises, and AI agents. This file indexes the markdown version of every page on www.min… (400 chars total)
• Also found /llms-full.txt (127,802 bytes). This isn't scored.

Not found. Checked /.well-known/mcp.json (404), /.well-known/mcp/server-card.json (404).

A server card lets agents see what your site offers without loading the page first.

{
  "name": "example-store",
  "version": "1.0.0",
  "description": "Search and browse the Example Store catalogue.",
  "tools": [
    {
      "name": "search_products",
      "description": "Search the catalogue and return matching products with prices."
    }
  ]
}

Not found. Checked /.well-known/agent-skills/index.json (404).

An agent skills index lists multi-step tasks agents can complete on your site. It's optional and worth the fewest points in this section.

{
  "skills": [
    {
      "name": "place-an-order",
      "description": "Search the catalogue, add to basket and complete checkout.",
      "path": "/.well-known/agent-skills/place-an-order.md"
    }
  ]
}

2 JSON-LD block(s) in the document.

• Sample: {"@context":"https://schema.org","@type":"WebSite","name":"Mintlify","creator":{"@type":"Organization","name":"Mintlify","url":"https://mintlify.com"}}
• Sample: {"@context":"https://schema.org","@graph":[{"@type":"Organization","@id":"https://www.mintlify.com/#organization","name":"Mintlify","url":"https://www.mintlify.com","logo":{"@type":"ImageObject","url":"https://mintcdn.co… (400 chars total)

robots.txt doesn't block GPTBot, ClaudeBot, Google-Extended, PerplexityBot.

• Explicitly addressed and allowed: GPTBot, ClaudeBot, Google-Extended, PerplexityBot

The page is served over HTTPS with a valid certificate.

• Connected to www.mintlify.com over HTTPS and the certificate is valid.
• Issuer: Let's Encrypt
• Valid until: Dec 11 15:10:55 2026 GMT

No origin trial token found, either in a meta tag or in the Origin-Trial header.

WebMCP is currently an origin trial in Chrome. Without a token for your domain, Chrome doesn't enable the API for your visitors, so your tools aren't available to them. They appear in this report because we simulate the API during the scan.

<!-- Option 1: meta tag, in <head> before any script that registers tools -->
<meta http-equiv="origin-trial" content="YOUR_TOKEN_HERE">

<!-- Option 2: response header (covers the whole response) -->
Origin-Trial: YOUR_TOKEN_HERE

<!-- Get a token: https://developer.chrome.com/origintrials -->

No WebMCP-related console errors, and no scripts blocked by your Content Security Policy.

• Your Content-Security-Policy header didn't block tool registration: worker-src * blob: data:; object-src data:; base-uri 'self'; upgrade-insecure-requests; frame-ancestors 'self' https://app.mintlify.com https://dashboard.mintli… (187 chars total)