https://isitagentready.com/check?url=https://yoursite.com
Paste that URL in a browser. In under ten seconds you get a score out of 100 and a list of exactly what is broken. This post explains what each signal means and what to fix first.
Why "agent-ready" is not the same as "SEO-friendly"
Search crawlers want indexable HTML. AI agents want something different:
A machine-readable summary of what the site does (
llms.txtat the root)A structured list of available endpoints or tools (
/.well-known/api-catalog)Auth instructions an agent can parse without reading a docs page (
auth.md)Clean semantic HTML with no token-wasting layout wrappers
A site can rank on page one of Google and still be completely opaque to a Claude or Codex agent trying to call it.
The free checker: isitagentready.com
Paste any URL into isitagentready.com and get a score out of 100 across four signal groups.
Signal group | What it checks | Max points |
|---|---|---|
Discoverability |
| 30 |
Readability | Semantic HTML, heading structure, alt text, no JS-only content | 25 |
API surface |
| 25 |
Auth legibility |
| 20 |
Scores under 50 mean an agent will either skip the site or hallucinate its capabilities. Scores above 80 mean the site can be wired into an agent workflow with minimal prompt engineering.
What each failing signal costs you
Missing llms.txt
Without it, a language model parsing your site has to infer structure from HTML — slow, token-heavy, and error-prone. Fix:
# /llms.txt
# yoursite.com
> One-line description of what this site does.
## Key pages
- [Home](https://yoursite.com/): main entry point
- [API docs](https://yoursite.com/api/): endpoint reference
- [Pricing](https://yoursite.com/pricing/): licensing terms
Drop that file at your domain root. The checker re-scores immediately.
Missing /.well-known/api-catalog
Agents discovering what a site can do look here first. A minimal catalog:
{
"apis": [
{
"id": "main",
"title": "Your API",
"baseUrl": "https://yoursite.com/api/",
"specUrl": "https://yoursite.com/openapi.json"
}
]
}
No auth signal
An agent that hits a 401 with no WWW-Authenticate header or auth.md file will stall. Add a header:
WWW-Authenticate: Bearer realm="https://yoursite.com/auth.md"
Then put human-readable and machine-readable auth instructions in /auth.md.
x402 for paid endpoints
If any endpoint requires payment, return a proper HTTP 402 with an x402 challenge rather than a 401 or an HTML paywall. An agent can read the challenge, discover the price, and settle on-chain — no human in the loop. Without it, the agent stops dead.
Reading your MCPize grade alongside the agent-readiness score
MCPize grades MCP servers specifically — it checks tool schema quality, transport compliance and error handling, graded A+ to B+. The agent-readiness checker covers the broader site surface: discovery, auth, and API legibility.
Run both. A server that grades A+ on MCPize but sits at 40/100 on isitagentready.com is reachable by a developer who already knows the endpoint URL, but invisible to an agent discovering it cold.
Fix order
Three things move the score the most, in this order:
Add
llms.txt— 15–20 point jump for most sites, takes under ten minutesAdd
/.well-known/api-catalog— unlocks the API surface score entirelyFix auth signalling —
auth.mdplus correctWWW-Authenticateheaders
JavaScript-rendered content, missing alt text, and thin heading structure matter, but are lower-leverage unless the site is a single-page app with no server-rendered HTML.
Next step
Run the checker on your own domain:
https://isitagentready.com/check?url=https://yourdomain.com
If the API surface score is zero, /.well-known/api-catalog is the fastest fix. If discoverability is the gap, start with llms.txt. The checker tells you exactly which line to address — no guessing required.