How we made our own site 100/100 on our own AI Access audit
Every file and setting on aisyndicate.com that earns the score, where to check it yourself, and the build check that stops us shipping if it slips.
On July 6, 2026, our homepage scored 100/100 with zero issues on the AI Access audit, the same automated scorer we sell. That is measured by our own audit, not an outside one. The score comes from open crawler access, an llms.txt and agents.md, Markdown copies of pages, structured data, and a build check that fails if those files slip.
What score did aisyndicate.com get, and who measured it?
Measured by our own AI Access audit, homepage, July 6, 2026: 100/100, zero issues. The result is published on our platform comparison page, next to five other AI-visibility and SEO platforms audited the same day with the same checks.
The AI Access audit asks one question: can AI engines reach, read, understand and trust this page? It groups its checks into five categories:
| Category | What it looks at | Our result |
|---|---|---|
| AI crawler access | Can AI crawlers reach and read the page? | Pass |
| Content and structure | Is the content laid out so a model can quote it? | Pass |
| Structured data (schema) | Does the page say what it is in code machines read? | Pass |
| Accessibility tree | Do screen readers and machine readers get a usable page outline? | Pass |
| Technical hygiene | The basics engines use to trust and credit a page | Pass |
How do we let AI crawlers in?
Crawlers are the robots AI companies send to read websites. Our robots.txt, the file that tells crawlers what they may read, names 56 crawlers one by one and allows every one of them, plus a catch-all rule for everyone else. The only thing blocked is /api/, which holds back-end endpoints, not pages. It also points to our sitemap.
The named list includes OpenAI's GPTBot, OAI-SearchBot and ChatGPT-User; Anthropic's ClaudeBot and Claude-SearchBot; Googlebot and Google-Extended; PerplexityBot; and Microsoft's bingbot.
In our experience, this matters more than anything else on the page. OpenAI says sites that opt out of OAI-SearchBot will not be shown in ChatGPT search answers. (OpenAI)
Two more settings on the homepage tell machines they may quote us: a robots meta tag with max-snippet:-1 (no limit on snippet length) and a tdm-reservation tag set to 0 (text and data mining allowed).
What is in our llms.txt, and why does it pass?
llms.txt is a plain Markdown file at the root of a site that gives AI tools a short, clean summary and links. The format was proposed at llmstxt.org: an H1 title, a short summary in a blockquote, then detail and lists of links.
Ours has the name as a heading, a one-line summary, a prose paragraph, an "Authoritative answers" block of question-and-answer pairs, our services and products with prices, a citation policy and a last-updated line. Our audit also checks length: an llms.txt should be 300 to 3,000 words. Ours sits just under that ceiling, on purpose.
That last number is the honest part of this story. Our llms.txt once grew to 3,606 words and started failing its own word-count check, on the audit we run for customers. We trimmed it twice in July 2026 (on July 17 and July 27) to get back inside the band. Long-form text now goes in llms-full.txt instead (see below).
What does agents.md do?
agents.md is written for AI agents, the tools that browse and act on a user's behalf rather than just read. Ours tells an agent what it can do on the site, which page answers which kind of question, how to tell us apart from unrelated companies with a similar name, and where the sibling files (robots.txt, llms.txt, sitemap) live.
Our audit checks it for an identity heading, an action section, a citation policy, links, references to sibling files, a last-updated line and a length of 80 to 1,500 words. Ours is 1,223 words and scores 100/100.
How do we stop the score from slipping?
A build check. The script scripts/check-ai-files.mjs scores our own llms.txt and agents.md with the same scoring code we use for customers. If either file scores below 100, the check fails.
It runs in two places: before every production build, and in the lint step that our code must pass before it can merge. So a bloated or broken llms.txt stops the release rather than going live. It also warns when llms.txt passes 2,900 words, so we trim before we hit the 3,000-word cap.
Why do our pages have Markdown twins?
Markdown is plain text with light formatting. It is much easier for a machine to read than a page full of code and layout. The llms.txt proposal suggests offering a clean Markdown copy of pages at the same address with .md on the end. (llmstxt.org)
We publish dozens of these. The homepage has /index.md, and pages such as /pricing/ and /methodology/ have their own. Almost every page links to its twin with a <link rel="alternate" type="text/markdown"> tag, and the homepage sends the same pointer in an HTTP header. Our audit checks for this link.
We also build llms-full.txt at deploy time: the full text of every HTML page in our sitemap, in one file, so an AI tool can read the whole site in one fetch.
What structured data is on the homepage?
Structured data (also called schema) is code that labels a page for machines: this is a company, this is its address, these are questions and answers. Our homepage declares Organization, WebSite, WebPage, FAQPage, Service, Offer, HowTo, ItemList, DefinedTerm, PostalAddress and ContactPoint types, plus a Speakable section marking the text best suited to voice answers. The Organization entry carries a disambiguatingDescription that says which "AI Syndicate" we are.
Pages are also served as ready-made HTML, not an empty shell that needs JavaScript to fill it in, so a crawler that doesn't run scripts still sees the words.
Do you need all of this to show up in AI answers?
No, and it would be wrong to say so. Google says there are no extra requirements to appear in AI Overviews or AI Mode, that you don't need new machine-readable files or AI text files, and that there is no special schema to add. A page needs to be indexed and eligible to show with a snippet. (Google Search Central)
What our score shows is narrower: that machines can reach and read every part of our homepage, and that we pass the bar we grade customers against. In our experience, the files that matter most are the unglamorous ones: a robots.txt that doesn't block the crawlers, and pages that say clearly who you are.
Want to see where your site stands? The AI visibility audit runs the same checks on yours.
Sources
- Overview of OpenAI crawlers — OpenAI. Read Sep 24, 2026.
- The /llms.txt file — llmstxt.org (Jeremy Howard). Read Sep 24, 2026.
- AI features and your website — Google Search Central. Read Sep 24, 2026.
Common questions
Is the 100/100 score independent?
No. It is measured by our own AI Access audit, the same one we sell. We publish it so you can check the files yourself, all of which are linked on this page.
When was the score measured?
The published score is from a homepage audit on July 6, 2026, shown on our comparison page. The build check re-scores our llms.txt and agents.md on every build.
Does a 100/100 score mean AI engines cite you?
No. It means machines can reach and read the page. Whether an engine names you depends on what it finds and trusts across the web, which is measured separately.
Can I copy what you did?
Yes. The files are public: /robots.txt, /llms.txt, /agents.md and /index.md. Start with robots.txt, because a crawler you block can't read anything else you publish.
Related pages.
See what AI says about your business.
Enter your domain and the free AI visibility check shows where ChatGPT, Gemini, Perplexity and Google's AI answers mention you — and where they don't.