← all toolsTool
Agent Readability Scanner
Score whether an AI agent can understand a page WITHOUT executing JavaScript. Fetches the URL once, parses the SSR HTML, and grades seven axes deterministically: text content, heading structure, metadata quality, internal links, schema presence, and a heuristic JS-dependency estimate. Returns an agent-perspective summary of what a bot reading raw HTML would actually understand.
Tool
Examples
Click an example to run the tool with pre-filled inputs.
Why this matters for AI agents
Most agent-visibility problems are SSR problems. A site that hydrates everything client-side ships an empty shell to GPTBot and ChatGPT-User; a site that puts the value prop, pricing, and CTAs in the initial HTML response ships a real page. The difference shows up in citation rate.
This tool runs in your browser’s tab and reports what the bot actually receives. No headless browser involved. The JS- dependency score is a heuristic from the SSR HTML, not a real render comparison. Flagged in the result.
Common fixes
Empty shell pageServer-render the hero, value props, and pricing. Treat JavaScript as enhancement; the page should be useful with JS disabled.
Missing or thin <title>Set a unique 30–60 char <title> per page. Format: "Page topic, Brand". Agents quote titles verbatim.
No JSON-LDAdd Organization + WebSite blocks site-wide; add page-kind schemas (Product, Article, FAQPage) per page. The
structured-data validator generates copyable templates.
Multiple H1sOne H1 per page. Multiple H1s confuse agents that pick the first as the topic; the rest are weighted as section headings.
Few internal linksCross-link related content from the body. Agents follow internal links to ground answers; thin internal linking caps citation breadth.
Linked rubric checks
- content.text_ratio
- content.js_required_for_primary_content
- content.headings_hierarchy
- markup.title
- markup.meta_description
FAQ
Why no headless browser?
Two reasons. First, the question this tool answers is "what does an agent reading the raw HTML response see?", and running a browser would mask that. Second, headless rendering per request would balloon latency and infrastructure cost. The full AgentSpeed scan does include a headless render pass; this tool focuses on the SSR-side signals.
How is JS dependency calculated?
A 0–1 heuristic. Penalties for: short SSR text content (<800 chars adds 0.4, <2,000 adds 0.15), low text-to-HTML ratio (<5% adds 0.4, <15% adds 0.2), and "needs JavaScript" / "Loading…" placeholder phrases in the body (adds 0.3). Capped at 1.0. Anything above 0.6 is treated as fail.
What does "agent perspective summary" mean?
A deterministic one-paragraph reconstruction of what an AI agent reading the SSR HTML would actually be able to summarise about the page. For thin pages, it admits the agent could not summarise the page; for content-rich pages, it reflects the title, description, and a snippet from the body.
Is this the same as the structured-data validator?
Different focus. The validator checks whether your JSON-LD parses and validates; this scanner checks whether agents can read your page at all. JSON-LD is one of several signals here, alongside text content, headings, and JS dependency.
Run a full agent-readiness scan
Full scanScore the entire site against 30 agent-readiness checks across discoverability, readability, structured data, actionability, and performance.
Get the full report →