Can AI agents read your docs?

Get an instant agent-readiness score for any documentation site — graded against the open AFDocs standard.

Free · No sign-up · Checks llms.txt, markdown, structure & more

Or install the Chrome extension to run on any page

What you'll get

A 0–100 score and letter grade, a category-by-category breakdown, and the specific checks to fix next. Here's a real result for docs.stripe.com.

Example AI Score result for docs.stripe.com — 84 out of 100, Grade B, passing 12 of 23 checks, with a category breakdown across content discoverability, Markdown availability, page size, structure, and URL stability.

What is an AI Score?

An AI Score measures how easily AI agents, assistants, and LLMs can read and use your documentation. AI tools increasingly answer questions and complete tasks by reading docs directly — but most documentation is built for human readers, with layouts, scripts, and navigation that machines can't parse. The result: agents miss content your readers can see.

The AI Score grades any site from 0 to 100 against the open AFDocs standard, so you can see exactly where agents succeed and where they get stuck — and what to fix first.

What the AI Score checks

The score is based on the open AFDocs standard — the things that make documentation readable by machines.

llms.txt

Does the site publish an llms.txt index — a sitemap for AI agents? It's the single highest-impact signal of agent-readiness.

Markdown access

Can each page be fetched as clean Markdown? Agents parse Markdown far more reliably than rendered HTML full of scripts and layout.

Content discoverability

Is real content near the top of the page, with meaningful headings — or buried below heroes, callouts, and navigation?

Page size

Are pages small enough to fit an agent's context window? Oversized pages get truncated, and the agent never sees the end.

URL stability

Are links durable and predictable, with proper redirects for moved pages instead of dead ends and 404s?

Reachability

Can an agent actually load the pages? Login walls and bot blocks keep AI tools out — and keep your docs out of their answers.

How it works

1

Paste a URL

Enter any public documentation site — your own or a competitor's. No sign-up, no install.

2

Run a fast scan

We check the page against the open AFDocs standard in a few seconds and grade it from 0 to 100.

3

See what to fix

Get a letter grade and a breakdown of exactly which checks pass and which to improve next.

AI Score FAQ

An AI Score is a 0–100 rating of how easily AI agents, assistants, and LLMs can read and use a documentation site. AI tools increasingly answer questions by reading docs directly, but most sites are built for human readers — so machines miss content people can see. The score grades a site against the open AFDocs standard and shows exactly which checks pass and which need work.
AFDocs is an open standard (afdocs.dev) for documentation that AI agents can read reliably. It defines concrete, testable checks — publishing an llms.txt index, serving pages as clean Markdown, keeping pages within agent context limits, using stable URLs, and more. The AI Score runs the AFDocs checks against your site and turns the results into a single grade.
llms.txt is a plain-text file at the root of a site that lists its documentation in a form AI agents can read quickly — a kind of sitemap for LLMs. It's one of the highest-impact things you can add for agent-readiness, and one of the first things the AI Score checks for.
Yes. You can score any public documentation site for free, with no sign-up required. Enter a URL and get an instant grade plus a breakdown of what to fix.
The tool runs a fast scan of the page you submit against the AFDocs checks, groups the results into categories — such as content discoverability, Markdown access, page size, and URL stability — and combines them into a 0–100 score and letter grade. It's a quick estimate based on a single page; a full build-time scan covers your whole site.
Start with the failing checks in your breakdown: publish an llms.txt index, make sure each page is available as clean Markdown, keep pages small enough to fit agent context windows, put real content near the top of each page, and use stable URLs with proper redirects. Docs built on Jamdesk pass most of these by default — llms.txt, Markdown export, and clean structure are generated automatically on every build.
It's designed for documentation sites — product docs, API references, knowledge bases, and guides. You can point it at any public http(s) URL, but the checks and guidance assume documentation content. Pages behind a login can't be read by agents, so they score low by design.
The URL of the page you choose to score is sent to Jamdesk's servers to run the scan, and is cached briefly (about 15 minutes) so repeat checks are fast. We don't track your browsing or store page contents. See the Jamdesk privacy policy for details.

Chrome extension

Score any site from your toolbar

Add the free Jamdesk AI Score extension and grade any documentation site as you browse — one click, no copy-paste. It only runs when you click.

Add to Chrome

Build docs that score well by default

Jamdesk generates llms.txt, Markdown export, and clean structure on every build — so your documentation is agent-ready out of the box, no checklist required. It also scores your docs against this same standard on every build, so you catch regressions the moment they ship. See how AI Score works in Jamdesk →

Start free trial

See how the major docs platforms scored in our agent-readiness benchmark.

We use cookies for analytics. Learn more