All scenarios/ Site owner

AI search readiness: which bots can read your site and what they could quote

Your customers ask ChatGPT, Perplexity or AI Overviews more often than they open ten links. You don't know whether those assistants can read your site at all: robots.txt may keep them out, a CDN may serve them a different page, and the text may only appear after JavaScript runs. You can check it from your own computer, without knowing SEO.

Who
A site owner who runs the site alone
Site
A company site, a blog or a service catalogue
What hurts
No idea whether AI assistants see the site
What's needed
A clear list: what is open, what is closed, what to fix
  1. Crawl the site and open the AI screen

    After a crawl, the AI screen shows a readiness score in four parts. Access and the text bots get are measured: robots.txt, sample pages fetched as each bot, text only after JavaScript, and nosnippet. Meaning and trust are the app's own estimate. The score never goes above the weaker of the two measured parts.

    AI — docs
    Molfar screenshot: The AI screen
    In the screenshot: The AI screen
  2. See who robots.txt lets in

    The robots.txt, bot by folder table marks every bot in every folder with ✓ or ✕, and hovering shows the rule and the line that matched. Closing a training bot is a legitimate choice: Google's AI Overviews and AI Mode use Googlebot, not Google-Extended.

    AI · robots.txt by folder — docs
    Molfar screenshot: The AI screen
    In the screenshot: The AI screen
  3. Check what the bots actually get

    Check as AI bots fetches about ten pages as a browser and with the User-Agent of every bot robots.txt allows, from your computer, without JavaScript. You see the status and the word count for the browser and for each bot. A CDN that verifies a bot's IP address may answer the real bot differently.

    AI · Check as AI bots — docs
    Molfar screenshot: The AI screen
    In the screenshot: The AI screen
  4. Find what blocks quoting

    Can AI read and quote it lists the problems with the number of pages each touches: measured ones first, such as nosnippet or max-snippet: 0, then style checks marked heuristic. Heuristics do not move the score, and each line opens its pages.

    AI · Can AI read and quote it — docs
    Molfar screenshot: Page passages
    In the screenshot: Page passages
  5. Check which questions the site answers

    Paste the questions your customers ask, one per line, and press Match questions. The embedding model on your computer compares each one with every section of the site and shows the closest. With Search Console connected, you also get queries phrased as questions, with impressions and clicks.

    AI · Which questions the site answers — docs
    Molfar screenshot: Semantic search
    In the screenshot: Semantic search
  6. Check llms.txt and Markdown for agents

    Molfar shows whether llms.txt and llms-full.txt exist, their sections and any links that lead to errors. For the sample pages, it shows whether the site answers Accept: text/markdown with Markdown and whether a .md version exists. No major AI search engine has said it reads llms.txt, and the app says so.

    AI · llms.txt — docs
    Molfar screenshot: The AI screen
    In the screenshot: The AI screen
  7. Save the reports

    The AI readiness and Page citability reports each answer one question: can bots fetch and read the site, and how might a retriever cut a page into passages. Both save as HTML or PDF.

    Reports — docs
    Molfar screenshot: Page passages
    In the screenshot: Page passages
01

An access map

Which folders robots.txt opens to which AI bots, with the rule and the line.

02

What bots get

The status and word count for the browser and for each bot on the sample pages.

03

A fix list

What stops pages from being quoted, with the number of pages, kept apart from style tips.

  • The readiness score is Molfar's own estimate, not a standard; only access and the text bots get are measured.
  • The check runs from your computer: a CDN that verifies the real bot's IP address may answer it differently.
  • Matching questions needs embeddings, which are built on your computer during the crawl.
If I block GPTBot, will my site disappear from ChatGPT?
Blocking a training bot is a separate choice. Bots that fetch pages to answer questions have other names, and Molfar shows each one separately.
Does a high score mean I will be quoted?
No. The score shows whether bots can read the site; what goes into an answer is up to the assistant.
Do I need llms.txt?
Molfar checks it if you have one, but no major AI search engine has said it reads it.

Try it on your own site.

Molfar is in preview and runs on Windows and on Macs with Apple silicon.

Get Molfar