HumonixCompanyGraph

HumonixCompanyGraph is Humonix’s website checker. It visits a company’s public website about once a month to see how ready the site is for search, AI answers, accessibility and AI agents, which common tools the page visibly loads, whether the company is hiring, and how its public security settings look.

What it readsrobots.txt first, then the homepage, sitemap.xml and llms.txt; the careers page the homepage links to and the public job board that page links to (at most four more requests, including the board’s public listing API); the clients or case-studies page the homepage links to (one request); public DNS records and one TLS handshake for the security grade. A careers page drawn by JavaScript may be opened once in a headless browser that loads no images, fills no forms, clicks nothing and follows no reloads. Nothing behind a login.
What it keepsScores from 0 to 100; the names of tools the page loads (for example an analytics or chat tag) and the social networks it links to; public job titles, their places and links; the names and web addresses of clients a company lists on its own site; an outside-in security grade. Not the pages’ text or images, and no personal information.
How oftenAbout once every 30 days per site, and once a week for companies with open roles; a few requests each time, a few sites at a time.
When it stopsWhen robots.txt disallows it (any rule naming a Humonix bot, or all robots), when robots.txt does not answer (a server error or a timeout: we try again later), and when a site shows a bot check or captcha: we never try to get past one.
How to opt outAdd these two lines to your robots.txt. They turn away every Humonix bot, not only this one:
User-agent: HumonixCompanyGraph
Disallow: /

Our other bots

HumonixDealRoomReads up to 8 public pages and 2 stylesheets of a company’s site, one request at a time, to prepare a private page for that company. Keeps short excerpts of public page text and the site’s colours for that page.
HumonixSiteCheckThe free site check, and the check behind a company’s private page: robots.txt, the homepage, sitemap.xml and llms.txt; to confirm an email address is published on a company’s own site, up to 4 of its contact or about pages.
HumonixQuickScanA paid scan a site’s owner ordered before 2026-11-15: up to 8 of that site’s public pages, once.
HumonixBotConfirms that a company’s web address is really theirs: robots.txt and one homepage request.

All of them follow the same robots.txt rules, including a site-wide “Disallow: /” for all robots. To have stored results removed, write to hello@humonix.com.