The checks
Everything Hide and Seek checks, each one something you can see for yourself.
40 checks in five categories. Each category scores on its own, and the score is their weighted mean; within a category, a check counts for its weight. A solid mark is something you can do; a hollow one is something others have to say about you. How the score is made.
Access 20 of 100
Can they get in?
- Search and answer bots are allowed30
Does robots.txt let in the bots that search and answer questions?
- Your firewall lets bots in30
Does the home page reach a bot the same way it reaches a browser?
- Pages can be indexed and quoted25
Do any pages tell search engines not to index them or not to quote them?
- One https address10
Does every version of the address lead to one https site, quickly?
- A fast home page5
Does the home page arrive in under two seconds?
- IndexNow5
Has the site published an IndexNow key, so it can tell search engines the moment a page changes?
- Content signalsreported, not scored
Does the site say what AI may do with its content, in robots.txt or an ai.txt?
Readability 15 of 100
Can they read it?
- Readable without JavaScript35
Is the text in the page itself, or does it only appear after JavaScript runs?
- Headings and landmarks25
Does each page have one main heading, headings in order, a main area, and a language?
- No walls before the content20
Does a cookie banner, age gate, or sign-up box come before the content in the page?
- Language and encoding10
Do the pages say what language they’re in and how their text is encoded?
- More text than markup10
How much of the home page is words, and how much is code?
- Images aren’t the text10
Is the home page’s message written in text, or only in pictures?
- Images have alt text10
Do the images in your content say what they show?
- A plain-text versionreported, not scored
Does the site offer its pages as Markdown or plain text, for machines that prefer it?
Identity 20 of 100
Do they know what you are?
- Structured identity30
Does the site say, in structured data, who or what it is?
- One name, everywhere25
Does the site call itself by the same name everywhere a machine looks?
- The identity’s details20
Does the site’s structured identity carry the details its type needs?
- Every page points home15
Does each page’s structured data say what the page is, and whose it is?
- An about page15
Is there a page that says who you are, linked from the home page?
- Title, description, and canonical10
Does each page have a title, a description, and a canonical link?
- Link previews10
When someone shares the home page, does the preview have a title, a description, and a picture?
- Articles have authors10
Do the site’s articles say who wrote them?
- Profiles that resolve10
Do the profiles the site claims as its own still exist, and do they name it?
Coverage 20 of 100
Is the whole site findable?
- Sitemap pages load35
Do the pages your sitemap lists actually load, with readable text?
- A sitemap25
Does the site have a valid sitemap, and does robots.txt say where it is?
- A feed15
Does the home page link to a working RSS or Atom feed?
- Recently updated15
Is there any sign the site has changed in the last few months?
- Missing pages are 404s10
When someone asks for a page that doesn’t exist, does the site say so?
- Articles are dated10
Does each article say when it was published?
- Linked from the home page10
Can the pages that matter be reached in one click from the home page?
- An llms.txt filereported, not scored
Does the site publish an llms.txt file?
Authority 25 of 100
Has anyone independent said you exist?
- A Wikidata item20
Is there a Wikidata item for the site’s name, with the site as its official website?
- Directory listings20
Is the site listed in the directories that matter for what it is?
- Pages cite their sources15
Do the site’s articles link to other sites as sources?
- In Common Crawl10
Is the site in Common Crawl, the open archive of the web that AI models learn from?
- Mentioned on Wikipedia10
Does any Wikipedia article link to the site, or carry its name as a title?
- Talked about elsewhere10
Has anyone on Hacker News posted or mentioned the site?
- Profiles link back10
Do the profiles your site claims link back to your site?
- On the web sincereported, not scored
When did the Internet Archive first save a copy of the site?
The score, from these checks
The Hide and Seek score is what a machine can check about a site: whether an AI system can get into it, read it, tell what it is, find all of it, and find it vouched for elsewhere. It is a number from 0 to 100 and a letter, the same for a free scan and a watched site, and every point of it is one of the 40 checks above.
Each category scores 0 to 100 on its own, as the weighted mean of its checks. The score is the weighted mean of the five categories by the weights beside their names. The letter is a plain banding: A from 90, B from 80, C from 70, D from 60, F below that. Only a true 100 shows as 100. A check that could not be run is skipped and left out of the maths, so it never counts for or against you; a check with no weight is reported and never counted.
What it isn’t
The score is not a ranking, not traffic, and not a judgment of what you’ve written. It doesn’t say whether an assistant names you when people ask; that is your North Star, which is counted beside it and is not a score. A 100 means one thing: nothing checkable is left undone. What remains after that is what your site says and who repeats it.
Who controls it
Nearly all of it is yours to do: a robots.txt that lets answer bots in, pages that read without JavaScript, JSON-LD that says what you are, a sitemap that is true. The report lists those under “Fix first”, worth the most first. A few checks have to be earned from others (In Common Crawl, Mentioned on Wikipedia, and Talked about elsewhere). They count toward the score like any other, because they matter to an assistant, but the report marks them with a hollow chip and keeps them out of “Fix first”: there is nothing to fix, only work to be noticed for.