Built for the web agents are reading now

AI agents — crawlers, chatbots, autonomous tools acting on someone's behalf — are quickly becoming some of the most frequent readers of any given website. They don't see a site the way a person does: most don't run JavaScript, don't click around, and don't notice the things that make a site feel polished to a human visitor.

Almost every site on the web today was designed and built with only human visitors in mind. That gap — between what a site shows a browser and what it actually hands an agent — is mostly invisible unless you go looking for it.

We built AgentLitmus to make that gap visible. Point it at a URL and it scans the site the way an agent would: fetching the raw HTML, checking for the guides agents look for (like llms.txt and AGENTS.md), and flagging content that could mislead or manipulate an agent reading it. The result is a grade and a concrete list of fixes — not a black box.

AgentLitmus is an early-stage, actively developed project. The scanner, generators, and monitoring you see today are real and live — and the methodology behind every score is published openly on the methodology page so you can see exactly how a grade is calculated.

If you're interested in what we're building — as a user, a partner, or a contributor — we'd like to hear from you. See the contact page.