How does the AI-search readiness check work, and what can it not do?
You paste your home page source, your robots.txt and your sitemap, and one script in your browser runs a fixed set of checks on that text and builds the fix list and the files from it. It fetches nothing from your site, reads one page only, and cannot tell what any search engine or AI assistant currently shows. Automated AI agents wrote the script; no agent and no person builds a pack for an order.
This page describes what the script does, step by step, and where it stops.
What you give the script
- The HTML source of your home page, copied from "View page source" in your browser. Paste only the source of a site you own or are authorised to modify; you tick a box to confirm it before the check runs.
- Your robots.txt and your XML sitemap, pasted the same way. Both are optional. An empty field is read as "the site has none".
- Your site address. It is used only as text inside the generated files. The script does not open it.
- Your business details: name, category, address, phone and opening hours. They are written into the files exactly as typed. An empty field is left out of the files.
Nothing is fetched from your site, at any point. The script does not crawl, does not log in and does not send requests to your server. What you paste and type stays in your browser.
The free check and the paid pack
- Free, no key: the script runs every check below on the pasted text and shows a readiness score, the number of items found by priority and the first three items of the prioritised fix list in full.
- With a licence key: the page sends the key to this site to be checked. When it is accepted, the page loads the rule tables and file templates of the pack from this site and the script builds the files in your browser, from the same pasted text. You download them one by one or as a ZIP.
The score is 100 minus 15 points for each high priority item, 7 for each medium and 3 for each low, never below 0. It counts the checks of this script on the text you pasted. It is not a prediction of ranking, citations or traffic.
The checks
| Check | What the script looks at | What it reports |
|---|---|---|
| Text without JavaScript | The visible text in the pasted HTML, read without running scripts. | A page whose text is missing from the HTML until scripts run. |
| Noindex | The robots meta tag in the pasted HTML. | A page that asks crawlers not to index it. |
| robots.txt and AI crawlers | The rules that apply to OAI-SearchBot, PerplexityBot, Bingbot, GPTBot, ClaudeBot and Google-Extended. | Which of the six are blocked, quoting the line numbers that block them. Search crawlers and training crawlers are reported separately, and unblocking training crawlers is marked optional. |
| Structured data | Every JSON-LD block in the pasted HTML. | Blocks that do not parse; no Organization or LocalBusiness data; questions and answers on the page without FAQPage data. |
| Title | The title of the page and its length. | A missing title, or one outside roughly 15 to 60 characters. |
| Meta description | The meta description and its length. | A missing description, or one that is very short or over about 160 characters. |
| Headings | The heading outline of the page. | No main heading (H1), or an irregular outline. |
| Canonical tag | The canonical link of the page. | A missing canonical, or one that needs a second look, for example because it points to a different page. |
| Name, address, phone | Phone links, phone numbers in the text, addresses, and the business name in structured data and Open Graph tags, compared with the details you typed. | A name or phone on the page that differs from what you typed, or contact details that are not in the HTML at all. |
| Sitemap and robots.txt | The robots.txt and sitemap fields. | No sitemap, a sitemap that is not declared in robots.txt, or no robots.txt. |
Each failed check becomes one item in the fix list with a priority (P1 high, P2 medium, P3 low), the evidence quoted from your pasted text and the step to fix it. Checks that passed, and the things the script does not check, are listed at the end of the fix list.
How the files are built
- FIX-LIST.md holds every item, not only the first three, with a summary table at the top.
- llms.txt lists the home page, labelled with its main heading (or its title when there is no H1) and its meta description when it has one, then the addresses found in your pasted sitemap, then the phone and address you typed.
- robots-additions.txt adds one group per audited crawler that has no group of its own in your robots.txt, and repeats the path rules of your general group inside each one, so paths you keep private stay private. It adds no group for a crawler you have blocked in a dedicated group; the fix list tells you which line to remove if you want to unblock it.
- schema/organization.jsonld holds the business name, site address, phone, postal address and opening hours you typed, each only if given and readable. The schema.org type follows the category you chose.
- schema/faqpage.jsonld is built only when the page holds questions and answers and has no FAQPage data. The questions and answers are copied from the page.
- meta-heading-fixes.csv has one row each for the title, meta description, H1, heading outline and canonical tag: the current value, its length, a proposed fix built from the page's own text and the details you typed, and a note saying where the proposal came from. When there is nothing to build from, the row says so and leaves the proposal empty.
- README.md says where each file goes and how to install it, and lists the details that were written into the files.
The sample is an example generated from a demo site.
Limits
- No results are promised. You receive files and a prioritised fix list. Whether search engines or AI assistants then rank, cite or send visitors to your site is not something the pack can control or guarantee.
- It is software, not a consultant. No person reads your page, your result or your files, and there is nobody to tailor the pack to your business.
- It reads one page. Other pages of the site are not covered, and neither are duplicate titles or descriptions across pages.
- It sees only what you paste. It does not see HTTP response headers, redirects or status codes, and it cannot tell whether an llms.txt file already exists on your site. If you paste the rendered DOM from developer tools instead of the page source, the result describes something crawlers do not receive.
- It does not run JavaScript. That is on purpose, because it shows what a crawler that does not run scripts gets. For a page that builds all of its content with JavaScript, the pack will mostly report that problem and will have little text to build the files from.
- It does not ask AI assistants about your business. It checks your page, not what any assistant currently answers.
- Address detection is limited to structured data, HTML address elements and street addresses written in the United States format.
- The files are drafts built from your own words. The script does not write new marketing copy.
- llms.txt is a proposed convention. The major AI providers have not confirmed that their crawlers use it.
- It does not install anything. You or the person who maintains your site put the files in place.
- Nothing is stored. The files exist only in the open page. Download them before closing it; to get them again, paste the page and enter the key again.
- Verification status. The generator's core functions are run against four fixed test cases by a script each time the product and the site are published. Each time the free check page loads, it runs its own self-test and hides the buy link if the generator fails in your browser. That script does not run the generator page in a browser, the licence key step or any real site; the four test cases are on the test fixtures and verification status page.
Delivery and refund
Delivery: instant, in your browser. The files are generated by the script as soon as the licence key is accepted, and you download them from the page. Nothing is emailed.
Refund policy: full refund on request within 7 days of purchase. No reason is needed. The refund covers the case where the key does not unlock the pack and the case where the pack cannot be produced for your site. How to request a refund: refunds are requested through Gumroad, by replying to the Gumroad receipt email for your order within 7 days of purchase. Refunds are made through Gumroad to the original payment method.
Price
The licence key costs $149, one payment, no subscription. Run the free check first; the buy link is under Price, delivery and refund on the home page.