Machine reading.
Machine reading fetches an address from this site’s own server the way a crawler would and shows, line by line, what the page tells a machine.
What is read is the page’s markup: title and description, canonical address, language and hreflang set, JSON-LD schema types, what robots.txt tells AI agents, and whether llms.txt and a sitemap exist. The tool does not interpret the page the way a person would; it writes down what is in the markup, line by line, and leaves a dash where nothing is.
Why from the server
A browser cannot read another site on its own: the content security policy and CORS prevent it. So the address goes to this site’s own server; the server fetches it under its own identity (the request header carries the tool’s name and address), returns the result to your browser and stores neither the address nor the result. That is what sets it apart from the other tools on this site, and the privacy statement says so separately.
What the lines say
The first line is the request itself: the address reached, the status code and the time taken. If there is a redirect chain, at most four hops are followed and the final address is written. The schema line lists the types read from JSON-LD blocks; a block that cannot be parsed is marked as such. The hreflang line carries the language codes the page declares. The robots.txt line gives, agent by agent, what the file says to AI agents: if the agent has its own group that group applies, otherwise the general group; no rule at all is also an answer.
The llms.txt line measures presence only; a file that returns HTML does not count. No effect is claimed: the file is a proposal, and this site’s observatory measures its adoption separately. How it differs from robots.txt and the layer as a whole are written separately.
Its limits
The tool reads public addresses only: local addresses, IP numbers and this site itself are refused. The body is cut at 400 KB, the request gives up after six seconds, at most four redirects are followed, and reads are limited to twelve per minute per IP address. No JavaScript is executed. A site that builds its content on the client may look empty here, which is exactly what the machine sees.
How to use
- Type the address of the page you want to read
- Press Read; the server fetches the page, robots.txt, llms.txt and the sitemap
- Read the lines: each one is something that is really in the markup, not an interpretation
This tool sends the address you enter to this site’s own server; the server reads that address, returns the result to your browser and stores nothing.