# AIFindUs

> AEO/GEO audit for any URL. AIFindUs measures whether AI answer engines can actually
> read, quote and attribute a website — by fetching each page twice (with and without
> JavaScript) and probing every AI crawler user-agent by HTTP status code — then
> generates a deployable artifact bundle grounded in the site's own content.

- Canonical site: https://csqueezy.com
- Category: DeveloperApplication / SEO and AEO tooling
- Primary users: developers, technical SEOs, growth engineers

## What it measures

- Render parity: the difference between a no-JavaScript fetch and a rendered one.
  Most AI crawlers do not execute JavaScript, so that gap is content no answer
  engine can cite.
- Verified crawler access: per-bot HTTP status, distinguishing an edge/WAF block
  from a robots.txt rule. A robots.txt Allow line plus a 403 is a silent failure.
- Passage structure: whether content is retrievable as passages and quotable as
  self-contained claims.
- Entity consistency, structured data validity, freshness, and off-site
  corroboration.

## How scoring works

Scores are computed in code, not by a language model, so the same evidence always
produces the same number. Weights: render parity 25, verified access 20, passage
structure 20, entity consistency 15, structured data 10, freshness 5, off-site 5.
Off-site presence is also reported separately, because it is the larger lever for
citation and on-site wins should not dilute it.

## Grounding

Generated artifacts are built from a fact-sheet verified against crawled text. Any
claim that cannot be traced to observed content is dropped rather than published.
Compliance and SLA claims are deliberately withheld from structured data until a
human confirms them.

## Docs

- [Architecture and evidence review](https://csqueezy.com/docs/plan): the nine-block design and
  the research the weights are based on

## Optional

- [Full text mirror](https://csqueezy.com/llms-full.txt)

# Note on this file: no major answer engine commits to reading llms.txt, and Google
# states its Search and AI surfaces do not use it. Ahrefs checked 137,210 domains in
# May 2026: 28% publish one, and 97% of those published files were never requested
# that month. The same study found the agents that DO fetch it are agentic and coding
# tools — Claude-Code, statespace-indexer — which arrive because someone pointed them
# at a URL rather than by crawling for you. So this file has a real but narrow
# audience. AIFindUs scores it zero FOR CITATION, which is a statement about answer
# engines and not a claim that the file is useless.
#
# It is also treated as code. Ahrefs documented bad actors probing llms.txt for prompt
# injection, so nothing here is written to read as an instruction to a model.