mathei.com

Return a markdown version of HTML pages when agents request it via Accept: text/markdown so LLMs ingest your content cleanly.

No markdown response for Accept: text/markdown

Enable Markdown for Agents so requests with `Accept: text/markdown` return a markdown version of your HTML response while HTML stays the default for browsers. Set `Content-Type: text/markdown` on the markdown variant and add `Vary: Accept` so caches differentiate the two representations. Verify with `curl -H "Accept: text/markdown" https://yoursite.com/`.

Publish an A2A Agent Card so other agents can discover your agent-to-agent capabilities and supported interfaces.

A2A Agent Card is missing

Publish /.well-known/agent-card.json describing your agent: `name`, `description`, `endpoints`, `version`, and a complete `supportedInterfaces` array with the protocols and message types you accept. Include a `capabilities` block so calling agents can negotiate, and keep the URL stable across deploys — agents cache discovery responses.

Publish a discoverable agent-skills index so AI agents can find and load your structured skill bundles.

Agent Skills index is missing

Publish a skills discovery index at /.well-known/agent-skills/index.json (per the Agent Skills Discovery RFC v0.2.0) with a `$schema` field and a `skills` array. Each entry should have `name`, `type`, `description`, `url`, and a `sha256` digest of the SKILL.md so agents can verify integrity. Make sure each `url` resolves to a reachable SKILL.md.

Publish an API catalog at /.well-known/api-catalog so agents can discover your API documentation and schemas without crawling.

API catalog is missing

Create /.well-known/api-catalog returning `application/linkset+json` with a `linkset` array. Each entry should include an `anchor` URL for the API and link relations like `service-desc` (OpenAPI spec), `service-doc` (human docs), and `status` (health endpoint). See RFC 9727 Appendix A for examples and RFC 9264 for the linkset format.

Serve an MCP Server Card so agents can discover your Model Context Protocol surface and connect with the right capabilities.

MCP server card is missing

Serve an MCP Server Card at /.well-known/mcp/server-card.json with `serverInfo` (name, version), the transport endpoint URL, and the `capabilities` you support. Accept /.well-known/mcp.json as a fallback for clients on the legacy path. The schema is being standardized — track the active proposal at github.com/modelcontextprotocol/modelcontextprotocol/pull/2127.

Expose OAuth/OIDC discovery metadata so agents can authenticate against your APIs programmatically.

OAuth or OIDC discovery metadata is missing or incomplete

If your site has protected APIs, publish /.well-known/openid-configuration (for OpenID Connect) or /.well-known/oauth-authorization-server (for OAuth 2.0) with `issuer`, `authorization_endpoint`, `token_endpoint`, `jwks_uri`, and `grant_types_supported`. This lets agents discover how to authenticate without parsing your docs.

Publish OAuth Protected Resource metadata so agents know which auth servers issue valid tokens for your APIs.

OAuth protected resource metadata is missing

Publish /.well-known/oauth-protected-resource (per RFC 9728) with your `resource` identifier, `authorization_servers` (the list of OAuth/OIDC issuer URLs that can issue tokens for this resource), and `scopes_supported`. This tells agents which auth server to obtain access tokens from and which scopes to request.

Expose in-page tools to AI agents via the WebMCP browser API so agents can act on your site directly.

No WebMCP registration markers were detected

Implement the WebMCP API by calling `navigator.modelContext.provideContext()` with tool definitions exposing your site's key actions to AI agents. Each tool needs `name`, `description`, `inputSchema` (JSON Schema), and an `execute` callback. Register tools at the points an agent could meaningfully use them — after page hydration, on route changes, or when permission state changes.

Robots meta tags and X-Robots-Tag headers align with which pages should actually be crawled and ranked.

1 sampled public page(s) are marked noindex.

Audit `<meta name="robots">` tags and `X-Robots-Tag` response headers across the site. Remove `noindex` and `nofollow` from any page meant to appear in search, especially on staging templates that leaked into production. For pages that genuinely should be excluded (admin, internal tools), keep the directive but verify the URL isn't also linked from a sitemap.

• /cdn-cgi/content?id=8v0xVYYGXI5F27ca5gJy3KOrUmCckyoaymef2Osnbeg-1777297850.0121-1.0.1.1-JAoK5MdZIw8DhEStPPVUvBgJVbAzMODRptGXpjOKoaM

Every indexable page has a single meta description (50-180 chars) that previews the page accurately for search and social.

2 sampled page(s) have missing or multiple meta descriptions; 0 description quality warning(s) were detected.

Add one `<meta name="description">` tag to each important indexable page — 50–180 characters of distinct, human-readable copy that previews the page accurately. Avoid copying the title, padding with keywords, or duplicating descriptions across pages; search engines either rewrite duplicates or skip them entirely.

• mathei.com
• /cdn-cgi/content?id=8v0xVYYGXI5F27ca5gJy3KOrUmCckyoaymef2Osnbeg-1777297850.0121-1.0.1.1-JAoK5MdZIw8DhEStPPVUvBgJVbAzMODRptGXpjOKoaM

Maintain a sitemap of canonical URLs and reference it from robots.txt so crawlers find new and changed pages.

No valid sitemap was discovered

Generate /sitemap.xml listing your canonical URLs (split into a sitemap index if you exceed 50,000 URLs or 50 MB). Refresh it whenever you publish, retire, or rename a page, and reference it from /robots.txt with a Sitemap: directive so crawlers find it without guessing.

Every indexable page has exactly one descriptive, unique title that summarizes the page in 50-60 characters.

1 sampled page(s) have missing or multiple titles; 0 title quality warning(s) were detected.

Add exactly one `<title>` tag to every indexable page, ideally 50–60 characters. Make each title descriptive and unique across the site so search engines can disambiguate them. For sectional pages, include the brand as a suffix (`Page Topic | Brand`) so social previews still attribute correctly.

• mathei.com

Each page has a clear heading hierarchy with one h1 and properly nested h2/h3 sections so structure conveys meaning to readers and crawlers.

1 H1 issue(s) and 0 skipped heading hierarchy warning(s) were detected.

Use exactly one `<h1>` per page that conveys the page's subject, then nest `<h2>` sections beneath it and `<h3>` for sub-sections — never skip levels (h1 → h3) or wrap headings in `<div>` for styling. Headings communicate document structure to assistive tech and crawlers; write them as outline entries, not as styled text.

• mathei.com

Expose Link response headers so agents can discover related resources (API catalog, skills, agent card) without parsing HTML.

No discovery-oriented Link headers were detected on the homepage response

Add Link response headers on your homepage and key API endpoints pointing agents to discovery documents. For example: `Link: </.well-known/api-catalog>; rel="api-catalog"` for your API catalog, or `Link: </docs/api>; rel="service-doc"` for documentation. See RFC 8288 for the format and the IANA Link Relations registry for valid `rel` values.

Each indexable page declares one canonical URL pointing to itself or the definitive original, so search engines consolidate ranking signals.

0 canonical error(s) and 2 canonical warning(s) were detected in the sample.

Add exactly one `<link rel="canonical" href="...">` per indexable page in the document head, with an absolute URL pointing to the preferred version (after redirects, with the right protocol and trailing-slash policy). Self-canonical is fine for the original; cross-canonical only when consolidating duplicates. Never canonicalize to a noindexed or 404 page.

Internal links use descriptive anchor text and form a reachable graph so crawlers, readers, and AI agents can navigate context.

1 link crawlability or anchor-text warning(s) were detected.

Use real `<a href="...">` elements for navigation — never `<div onclick>` or button-styled spans, which crawlers can't follow. Write descriptive anchor text that summarizes the destination (avoid "click here", "read more", or naked URLs). Keep important pages within ~3 clicks of the homepage so crawl depth doesn't starve them.

• mathei.com

Each indexable page carries enough unique, substantive content to merit indexing on its own without overlapping siblings.

1 thin-content warning(s) and 0 duplicate-content fingerprint group(s) were detected.

Audit indexable pages for substantive, unique content — at minimum a few hundred words of original copy that delivers on the title and description. Consolidate near-duplicate pages with canonical tags or merge them into a single richer page. Don't ship placeholder "Coming soon" pages to production indexable URLs.

• mathei.com

The site declares a favicon (and ideally apple-touch-icon variants) via standard <link> tags so browsers, bookmarks, and previews show the right brand mark.

No favicon link was detected on the sampled homepage.

Add a `<link rel="icon" href="/favicon.ico">` (or a versioned `.png`/`.svg`) to the homepage head, plus an optional `<link rel="apple-touch-icon" href="/apple-touch-icon.png">` (180×180) for iOS bookmarks. Keep the URL stable across deploys so browsers and bookmarks don't lose the icon to cache busts.

Pages declare language and region alternates with hreflang annotations when content varies by locale, so search engines route the right version to the right reader.

No hreflang annotations were detected in the sampled pages.

On each localized page, add `<link rel="alternate" hreflang="...">` tags pointing to every language/region variant (including a self-reference). Use valid IANA language-region codes (e.g. `en-US`, `fr-FR`), absolute `href` URLs, and one `hreflang="x-default"` entry pointing to the version for unspecified locales. Each variant must reciprocally link back.

Sitemap entries match the live canonical URLs and stay consistent with what the site actually publishes.

No sitemap URL metadata was available from the already-fetched sitemap sample.

Make sure sitemap entries match what the site actually publishes: only canonical URLs, only indexable pages (no noindex), and `<lastmod>` values in valid W3C datetime format. Remove redirects, 404s, and non-canonical URLs from the sitemap so crawlers don't spend budget chasing them.

Pages expose Open Graph and Twitter Card metadata so they preview cleanly when shared on social, in chat, or by AI assistants.

No Open Graph or Twitter card metadata was detected in the sampled pages.

On shareable pages, add Open Graph and Twitter Card metadata in the head: `og:title`, `og:description`, `og:image` (1200×630 minimum), `og:url`, `og:type`, plus `twitter:card` set to `summary_large_image`. Make the OG content match the visible page so previews aren't misleading, and use a real image URL with stable dimensions so social platforms cache reliably.

Pages publish JSON-LD structured data so search engines and AI agents can extract entities, products, FAQs, and offers reliably.

No JSON-LD structured data was detected in the sampled pages.

Embed JSON-LD `<script type="application/ld+json">` blocks in the head describing the page's primary entity — Article, Product, FAQPage, Organization, etc. — using schema.org types and required fields per Google's structured data guidelines. Validate with the Rich Results Test before shipping, and keep the JSON strictly parseable (no trailing commas, no comments).

Content-Security-Policy is missing

Publish a baseline Content-Security-Policy for the site shell.

HTTP Observatory grade D with score 30.

Tighten response headers and HTTP posture where Observatory reports gaps.

HSTS is missing

Add a Strict-Transport-Security header on HTTPS responses.

Identify your own bot traffic with Web Bot Auth signed requests so receiving sites can verify and trust them.

Web Bot Auth HTTP Message Signatures directory is missing

Publish a JWKS at /.well-known/http-message-signatures-directory containing the public keys you use to sign outgoing bot requests, so receiving sites can verify those requests via HTTP Message Signatures (RFC 9421). Rotate keys on a regular cadence and keep the previous key in the JWKS until in-flight requests have aged out.

X-Content-Type-Options is not set to nosniff

Set X-Content-Type-Options: nosniff.

DNSKEY records were detected, but no DS delegation was observed.

No Cloudflare Challenge Page marker was detected on the homepage response.

Cloudflare reported cache status DYNAMIC.

PSI mobile performance score is 100.

Improve mobile Lighthouse performance issues highlighted by PageSpeed Insights.

Homepage Link headers do not advertise agent-useful discovery artifacts

Expose Link headers for api-catalog, service-desc, agent-skills, or related discovery artifacts.

Publish Agentic Commerce Protocol discovery so agents can transact with your commerce surfaces without scraping.

No ACP discovery markers were detected

Serve /.well-known/acp.json at the origin root with `protocol.name` set to `"acp"`, the `protocol.version`, `api_base_url`, supported transports, and `capabilities.services`. Agents can then discover your ACP implementation without first creating a checkout session.

Publish Machine Payment Protocol metadata so agents can discover paid endpoints in your API.

No MPP discovery markers were detected

Publish an OpenAPI document at /openapi.json with `x-payment-info` extensions on payable operations. Each operation should declare `intent` (charge or session), `method` (tempo, stripe, lightning, card), `amount`, and `currency`. Use the MPP SDK (`mppx` for TypeScript, `pympp` for Python) with framework middleware for Hono, Express, Next.js, or Elysia to handle the payment flow.

Expose Universal Commerce Protocol metadata so agents can transact for content or services through a standard interface.

No UCP discovery markers were detected

Serve /.well-known/ucp with your protocol version, declared `services`, supported `capabilities`, and reachable `endpoints`. Make sure any spec URLs and referenced JSON Schemas resolve so agents can validate against them at discovery time.

Support x402-style HTTP 402 payment flows so agents can pay for API access machine-to-machine.

No x402 discovery markers were detected

Add x402 payment middleware to your API routes so AI agents can pay for access via HTTP. Use `@x402/express`, `@x402/hono`, or `@x402/next` middleware with a facilitator URL and wallet address. Protected routes will return HTTP 402 with payment requirements that agents can fulfill automatically.

Scanned every discovered page (2).

No Cloudflare-generated error diagnostics were detected on the homepage response.

No Cloudflare product beacons were detected in the homepage response.

Detected 1 public DNS/service signal(s).