Free tool · Schema markup for AI search

Find out what AI engines actually see in your schema.

Paste a URL. Schema Sniffer reads the structured data on the page, finds the errors, the thin bits and the missed opportunities, and writes the whole thing up in plain English. About ten seconds, no signup, full report on screen.

How the sniff works

Three things, in about ten seconds.

No dashboard, no signup, no email until you ask for the snippets.

01 · Sniff

It picks up the scent

The tool fetches your page, pulls out every JSON-LD block, parses each one, and works out what kind of page it is looking at: homepage, article, product, service, contact, FAQ, case study, or something more bespoke.

02 · Read

It runs the rules

About twenty rules, grouped into errors (schema that is broken), flaws (schema that is present but thin) and opportunities (schema that should be there for that page type but is not). Every finding gets a severity and a stake.

03 · Speak

It tells you the story

Claude Haiku 4.5 rewrites the findings in plain English, explaining what each one costs you with the AI engines that actually use structured data. Then it generates the snippets you would paste in to fix them.

FAQ

Fair questions before you paste a URL.

What does Schema Sniffer actually check?

It fetches the URL you give it, extracts every JSON-LD block, parses each one, and runs it through about twenty rules covering errors, flaws and missed opportunities for the type of page it is looking at. The output is a score, a verdict, every finding written out in plain English, and paste-ready JSON-LD for the gaps.

Is it really free?

Yes, completely free. The full report shows up on screen with no email required. The only thing gated is the JSON-LD snippets that fix the missed opportunities, and that gate is one email field. You get the snippets on screen and a copy in your inbox.

Which AI engines does it think about?

The rules and the commentary are written for the engines that structured data actually feeds: ChatGPT, Claude, Gemini, Perplexity and Google's AI Overviews. It is not just a Rich Results validator with a fresh coat of paint. It is opinionated about what AI search wants.

Does it check microdata or RDFa too?

It detects them and flags their presence, but the rule engine focuses on JSON-LD because that is what every major AI engine reads most reliably. If you have microdata, migrating to JSON-LD is usually the recommendation anyway.

How many sniffs can I run?

Three per domain per day, twenty per IP per day. Enough to check the most important pages without burning the API bill it costs to keep this free.

Will it work on a JavaScript-rendered site?

If your structured data is rendered server-side or baked into the HTML at build time, yes. If your schema is injected by client-side JavaScript only, version one will miss it. That is increasingly rare and is on the list to address.