For AI agents and crawlers

Your site can look fine and still read as empty.

ChatGPT, Claude and Perplexity read your site for people who never visit it, and some of them get nothing. See what they reach, what they miss, and what to change.

100+ checks, asked as ChatGPT, Claude, Gemini, Perplexity, Copilot, Meta AI, Apple Intelligence, Alexa and other crawlers. The request and response behind every finding. Free, no account needed.

What stops an agent never shows in your own browser:

  • A line in robots.txt tells it to stay out.

  • A bot challenge turns it away before it reads a word.

  • Text that needs JavaScript never arrives, so the page reads as empty.

Learn more

The blocking map

See which pages each agent cannot read.

This is a real map of bbc.com, made on 29 October 2026. Choose an agent, then choose a template to see the pages and the reason.

Scan a page, then map its site from the report. 50 pages, no account.

How it works

One page, then the whole site, then every week.

  1. One page

    What to change, in order

    100+ checks on any page. Anything that blocks agents comes first, then the fixes that cost them most, each with code to paste and where it goes.

    Scan a page
  2. The whole site

    Every template, not just the homepage

    Read the sitemap, group its pages by template, and scan one of each. Pages and Issues show what is wrong where, across all of them.

    Extract a site's pages
  3. Every week

    Hear when something slips

    A site you watch is scanned again each week. You get an email when a check that passed starts failing. This is the part that needs an account.

    Create an account

The proof

Every finding shows the request, the response and the line.

We read each page as a browser, as GPTBot, ClaudeBot and PerplexityBot, and in a real headless browser. Your developer can check every finding.

What we check

100+ checks across 10 categories.

Checks that do not apply to your kind of site — commerce on a documentation site — are marked not applicable, never failed.

Checks per scan, against the 29 tools in this space
  • Median tool13
  • Deepest tool46
  • Braille100+
  • Content accessibility & extractability16
  • Crawler access & permissions20
  • Discovery surfaces12
  • Structured data & metadata10
  • Agent capability & protocol exposure15
  • Identity, trust & auth15
  • Accessibility (WCAG)14
  • Agentic commerce & payments13
  • Freshness6
  • Performance & reliability11
Show what each category covers
  • Content accessibility & extractability

    16

    Whether the substance survives the trip to a non-browser client.

    • C1Primary content server-rendered in raw HTML
    • C2Static vs. rendered content parity
    • C3Markdown content negotiation
    • C4Per-page markdown twin
    • C5Semantic HTML & heading hierarchy
    • C6No login, paywall or consent wall on core content
  • Crawler access & permissions

    20

    Can an AI bot legally and physically reach the page at all.

    • A1robots.txt exists and parses
    • A2Named AI-bot rules in robots.txt
    • A3No blanket Disallow: / against agents
    • A4Content Signals / AI-preference directives
    • A5meta robots & X-Robots-Tag indexability
    • A6Bot wall, CAPTCHA, WAF or CDN block
  • Discovery surfaces

    12

    The files and headers that tell an agent what exists and where.

    • B1sitemap.xml present, valid, declared
    • B2llms.txt present
    • B3llms.txt structural validity
  • Structured data & metadata

    10

    Machine-readable meaning layered over the HTML.

    • D1JSON-LD / schema.org present and valid
    • D2High-value schema types
    • D4Title & meta description quality
  • Agent capability & protocol exposure

    15

    What an agent can do here, not just read.

    • E1MCP server endpoint
    • E2MCP Server Card
    • E3WebMCP
  • Identity, trust & auth

    15

    Who the agent is, who you are, and why either should be believed.

    • F1Web Bot Auth / HTTP Message Signatures
    • F2OAuth / OIDC discovery
    • F3OAuth Protected Resource metadata
  • Accessibility (WCAG)

    14

    Human-accessibility checks that double as machine-legibility checks.

    • J1Image alt-text coverage
    • J2Form labels & accessible control names
    • J3Valid ARIA and resolved references
  • Agentic commerce & payments

    13

    Machine-to-machine transaction rails.

    • G1x402 / HTTP 402 challenge
    • G2/.well-known/x402.json
    • G3Agentic Commerce Protocol (ACP)
  • Freshness

    6

    Signals that the content is current.

    • H1Visible published / updated dates
    • H2datePublished / dateModified in structured data
    • H3sitemap lastmod values
  • Performance & reliability

    11

    Speed and stability, measured from the bot's side.

    • I1Server response time to bots
    • I2Core Web Vitals (LCP, CLS, INP)
    • I3Page load under a hard threshold

Free

No account to start. One to keep watching.

Sign in later and everything you ran in this browser comes with you.

FeatureWithout an accountWith an account
Report with all 100+ checks10 pages a day100 pages a day
Run one check on its own50 checks a day300 checks a day
Extract a site's pages and scan several5 sites a day20 sites a day
Home, Pages, Issues, HistoryIn this browserOn any device
Blocking map50 pages from the sitemap, 1 a day150 pages, from the sitemap or by following links
Weekly scan and an email when something slipsNeeds an address to write toUp to 5 sites

See what agents get from your site, and what they miss.

100+ checks, asked as ChatGPT, Claude, Gemini, Perplexity, Copilot, Meta AI, Apple Intelligence, Alexa and other crawlers. The request and response behind every finding. Free, no account needed.