Every check.

The checks behind both scores, from the same check registry the scanner and the command line tool use, so this list changes when they do.

Each check links to what it measures, why it matters and the evidence behind it. In a terminal, lensy explain <id> prints the same text. How Lensy scores a page explains how the checks add up.

AI search

Six checks worth 100 points between them. Each check gives full, partial or no points.

Coding agents

Each check earns a share of its weight (critical 10, high 7, medium 4, low 2; a warning earns part of it). The score is the weight earned over the weight of the checks that ran, out of 100. Some critical failures cap the score: with no llms.txt it is at most 59.

CheckWeight
llms.txt exists10
llms.txt format4
llms.txt size7
llms.txt links resolve7
llms.txt links point to Markdown7
Page points agents to llms.txt (HTML)7
Page points agents to llms.txt (Markdown)4
.md version of the page7
Markdown on request4
Content without JavaScript10
Markdown size7
HTML size7
Download size4
Content starts near the top4
Complete in one fetch4
Tabbed content readable4
Headers inside tabs2
Code fences closed4
Links work outside the site4
Embedded data size4
Real 404s for missing pages7
Redirects4
llms.txt covers the sitemap4
Markdown matches HTML4
Cache headers2
Public without login10
Alternative access to gated docs4
Bot protection lets agents in7
Headings an agent can navigate4

Coding agents without AFDocs

If AFDocs can't run on a page, the Coding agents score comes from these six Lensy checks instead, worth 100 points between them.