Can an AI agent
act on your site?
Enter your URL. The checker fetches your page, probes /.well-known/ for agent manifests, reads your forms, pricing markup, and robots policy, and scores whether an AI agent can complete tasks on your site — not just read it, but act on it.
What agent readiness means
Search engines read your site. AI agents need to act on it — find pricing, fill a form, discover an API, complete a purchase. The gap between readable and actionable is where most sites fail AI agents without knowing it. This checker runs the same six deterministic checks the IndexHalo report uses and tells you exactly where the blockers are.
The score is not a vibe check. Each of the six checks is deterministic: the agent manifest either exists at /.well-known/ or it doesn't; the forms either have labelled inputs or they don't; the pricing either carries Product/Offer schema or it doesn't. Every check reports what it found, not just pass or fail.
The agents.json opportunity
There is no ratified standard for declaring what an AI agent can do on a site — yet. The proposed convention (agents.json at /.well-known/) is early-mover territory. Publishing one before the standard settles means you own the query space: when agents look for a manifest, yours is there. When the checker finds no manifest, it generates a draft from your page's real forms — the actions, capabilities, and policies your site already exposes — ready to publish.
What this does and does not measure
Agent readiness is measured — the same checks every time, over your live page, with no model in the loop. What it does not do is test whether an agent can actually complete a multi-step task end to end (that needs a live agent run, which is part of the paid platform). This checker is the free validator that tells you whether the building blocks are in place.
Frequently asked questions
What is agent readiness?+
Agent readiness measures whether an AI agent can complete tasks on your site — not just read it, but act on it. Can it find your pricing? Fill your contact form? Discover your API? Complete a purchase path? Each is a deterministic check: the agent either can or cannot, and the score tells you where the blockers are.
What does the checker look for?+
Six things: an agent action manifest (agents.json, MCP, or ai-plugin.json at /.well-known/), llms.txt guidance for AI agents, form semantics (labelled inputs, CAPTCHA detection), structured pricing and availability markup (Product/Offer schema), whether on-demand agent fetchers are allowed in robots.txt, and API discoverability (OpenAPI or docs-rich llms.txt).
What is agents.json?+
A proposed standard for declaring the actions an AI agent can take on a site — find pricing, book a demo, place an order, call an API. There is no ratified standard yet, which is exactly why publishing one early is valuable: you own the query space before it settles. When the checker finds no manifest, it generates a draft from your page's real forms.
Why does CAPTCHA lower the score?+
A CAPTCHA is a hard block for an AI agent — it cannot solve it, so any task behind a CAPTCHA-gated form is unreachable. The checker detects known widget signatures (reCAPTCHA, hCaptcha, Cloudflare Turnstile) in the markup, not prose mentions, and flags the form as agent-blocked.
Is anything stored?+
The checker fetches your live page, robots.txt, llms.txt, and probes /.well-known/ for manifest files. The analysis runs in the request and returns the score. Nothing is stored beyond the standard rate-limit counter.