Certificate of measurement No. WOA-2026-1022-0417 Page 1 of 4

Can AI assistants find, read and recommend your site?

Recommendations from ChatGPT, Claude and Gemini are the new word of mouth. Word of Agent sends the requests that search engines and AI assistants send, records what your site answers, and says what to change. Every reading keeps its evidence and a command that proves the fix.

Typing a domain makes the certificate out to it. Then sign up with Google or your email and the real audit runs.

Method: 14 crawlers × every page in the sitemap; 4 assistants × several runs per question
Traceable to: the raw responses, kept with each finding
Issued: 22 October 2026

Readings · sample
Measured	Reading
Crawlers let in	11 of 14
Pages readable without JavaScript	38 of 41
Named by ChatGPT	4 of 12
Named by Claude	7 of 12
Cited by Gemini	2 of 12
Wikidata item	None

Sample readings. Your certificate is issued when the audit runs.

Schedule A No. WOA-2026-1022-0417 Page 2 of 4

What is measured

Twelve checks, each answering a question a business owner would ask. Crawler checks run on every page in the sitemap, up to 500 by default, and the report says when it sampled.

Check	The question it answers	Measured against
Can they find it?
Search engines	Are the pages in Google and Bing, and are they found?	Search Console and Bing Webmaster Tools
Crawler access	Can search engines and AI crawlers get in?	robots.txt per crawler, and the server's answer to each crawler's user agent
HTTPS and hosts	Does the site answer at one secure address?	Redirects, HSTS, one host
Can they read it?
Readability	Can crawlers tell what each page is?	Titles, descriptions, canonicals, headings, structured data
Internal links	Can crawlers get around?	Broken, redirecting and orphan pages, click depth
Rendering	Do crawlers that skip JavaScript see the same page?	Headless Chrome against the raw HTML
Performance	Is the site fast for visitors and for AI tools fetching it live?	PageSpeed field and lab data, server response times
Appearance	How does the site look in search results, link previews and assistants?	Favicon and site name for Google, preview images and titles, snippet controls
Rich results	Can search and assistants show your products, business and articles in full?	Product price and availability, logo, breadcrumbs, address and hours, article dates
Agent readiness	Can an AI agent find prices, policies and who you are?	Prices as text, trust pages, llms.txt, Markdown
Will they recommend it?
AI answers	Do AI assistants name, cite and recommend you when customers ask?	ChatGPT, Claude and Gemini with web search, several runs per question
Presence	Does the wider web identify the business?	Wikidata, Wikipedia, linked profiles

How readings are graded

Fail: Something stops search engines or AI assistants from finding, reading or recommending the site today. Fix it first.
Warning: It costs visibility, now or later. Blocking AI training crawlers is a valid choice: it is reported here, never fixed unasked.
Info: Worth knowing. Nothing to change yet.
Pass: Measured and fine.

Schedule B No. WOA-2026-1022-0417 Page 3 of 4

A reading in full

Every finding has the same parts: the result, the evidence, the steps to confirm the cause, the fix, and a command that proves it. This one is a sample for your site.

access.http.blocked Fail

OAI-SearchBot is refused by the server on 8 of 8 pages tested

Result: OAI-SearchBot feeds ChatGPT search, so blocked pages cannot appear there. Requests with its user agent got a bot challenge (HTTP 403), while a browser got the page. The site is served through Cloudflare.
Evidence: Sent as Request Answer

OAI-SearchBot GET /pricing 403, bot challenge

OAI-SearchBot GET /about 403, bot challenge

Browser GET /pricing 200
Diagnose: In the Cloudflare dashboard, open Security > Settings (Bots) and check whether "Block AI bots" is on.

Open AI Crawl Control and check whether OAI-SearchBot is set to Block.

Open Security > Events, filter by user agent containing "OAI-SearchBot", and see which rule acted.
Fix: Allow OAI-SearchBot in Cloudflare's bot settings.
Verify: woa audit yoursite.com --only access

Sent as	Request	Answer
OAI-SearchBot	`GET /pricing`	403, bot challenge
OAI-SearchBot	`GET /about`	403, bot challenge
Browser	`GET /pricing`	200

Schedule C No. WOA-2026-1022-0417 Page 4 of 4

For agents

The CLI, the API and the MCP server return the same JSON; the dashboard is one view of it. Finding ids are stable, so an agent can branch on them, fix what it can reach, and run the verify command itself.

$ woa audit yoursite.com \
    | jq '.checks[].findings[]
          | select(.severity == "fail") | .id'
"access.http.blocked"
"agent.pricing.none"
"render.jsonld-needs-js"

Connect it to Claude Code, Cursor or any MCP client with woa mcp. Tools include audit_site, get_report and get_finding.

Your keys

Asking the assistants costs money per answer. On the free plan you bring your own keys for ChatGPT, Claude and Gemini. Keys are encrypted at rest, shown back only as their last 4 characters, and never logged. Agents can see whether a key is set, never read or change it.

Crawler requests come from our network with each crawler's user agent, not from the crawler's own network. CDNs that check crawlers by IP can treat the real crawler differently, and every report says so.

Issue a certificate for your site

Sign up with Google or your email; your first audit runs as soon as the site is added. Already have an account? Sign in. Or run it from a terminal:

woa audit yoursite.com