AI readiness
Can AI crawlers and agents reach and read your site? Free — no usage balance used.
Technical SEO → AI readiness checks the homepage of the project's primary domain and related resources to answer one question: can AI crawlers and agents reach and read your site?
It is the precondition of AI visibility. AI visibility measures the result — whether AI answers mention you. Readiness measures whether AI can read you at all. A high score means AI can reach and read the site, not that AI will cite it.
Free
The check runs on our own servers and calls no paid API, so it is free on every account, subscribed or not, and uses no usage balance.
| Limit | Value |
|---|---|
| Per organization per UTC day | 20 checks |
| Per organization at a time | 1 check |
| Same site again within 10 minutes | the previous result is returned |
What it checks
Core checks (scored, out of 100)
| ID | Check |
|---|---|
| F01 | Public access, HTTPS & redirects |
| F02 | robots.txt & crawl rules |
| F04 | Indexing & snippet limits |
| F05 | Sitemap discovery & validity |
| F06 | Canonical URL |
| F07 | Page language & hreflang |
| F08 | Title & meta description |
| F09 | Semantic & navigable structure |
| F10 | Content without JavaScript |
| F14 | Link discoverability |
| F17 | Content size & truncation risk |
Bonus (up to +10, absence never subtracts)
| ID | Check |
|---|---|
| F03 | AI crawler declaration matrix |
| F11 | Structured data (JSON-LD) |
| F12 | Structured data matches visible content |
| F13 | Organization, author & source signals |
| F15 | Markdown discovery & content negotiation |
| F16 | llms.txt / llms-full.txt spec and link sample |
| F18 | Agent, API & service discovery |
F03 judges each AI crawler by its exactly named User-agent group in
robots.txt, split into training, search and user-triggered crawlers; the *
group is not a statement about AI crawlers.
Agent protocols (G01–G15, not scored)
Which interfaces the site offers AI agents: a Markdown alternate, an API catalog, OAuth metadata, an MCP server card, Agent Skills, DNS discovery records and more.
Information items (H01–H04, not scored)
Share preview, feed, site icon, machine-readable date.
Reading the result
The report reads verdict → how to fix → details. Each item has one of these states:
| State | Meaning |
|---|---|
| Failed / Partial | Reliable evidence shows a problem; fix steps included |
| Declared · broken | The site declares the capability but it is invalid |
| Optional · not found | Optional capability; whether to add it depends on the site; never subtracts |
| Could not verify | The evidence could not be trusted or obtained (encoding error, request limited). Neither a pass nor a defect |
An incomplete check shows no score — for example when the homepage could not be fetched, is not HTML, robots.txt is unreachable, or any scored check could not be verified. The page then only explains why.
In history, a score change is shown only between two complete checks of the same target and scoring caliber; otherwise it reads "different caliber — not comparable".
How the crawler works
The check is made by the AIEO-Readiness crawler. It obeys robots.txt, sends
GET requests only, uses ports 80 and 443 only and runs no JavaScript — at most
80 requests and 90 seconds per check. The full request list, rate limits and
how to block it are on /bot.
If your robots.txt disallows AIEO-Readiness, the check says so plainly; it
never works around a block.
Reading it over MCP
The read-only MCP tool get_ai_readiness_report returns the latest report's
status, scores and items to fix. It only reads stored reports and never
starts a check.
Retention
Each project keeps the details of its latest 50 reports, for at most 180 days. Older reports keep their scores only, for one year.