AI Agent Readiness Audit

Sunrise Plumbing (sample) · https://www.example.com · October 15, 2026

58/100Readiness score

first runChange

12Pages read

9Fixes listed

The short version

Your AI agent readiness score is 58/100. The biggest wins: let AI search and answer engines read your site (+15.9); publish an llms.txt file (+8); business hours as text (+4).

Scores by category

Category	Score	Weight
AI crawler access	32/100	25%
Facts readable as text	73/100	22%
Structured data	70/100	20%
Actionability for agents	67/100	15%
Basics	100/100	10%
llms.txt	0/100	8%

Fix list (biggest gains first)

Fix	Why it matters	Gain
Remove the robots.txt rules that block PerplexityBot (see robots-ai-section.txt).	AI search and answer engines can only quote, cite or recommend pages they are allowed to read.	+15.9
Upload the drafted llms.txt to the root of your site.	llms.txt is a proposed Markdown file (llmstxt.org) that gives AI tools a short, clean map of your business and key pages. Adoption by AI companies varies; it is cheap to add.	+8
Write your hours as text (for example 'Mon-Fri 7am-6pm') on the contact page and footer.	'Are they open now?' is one of the most common questions assistants answer.	+4
Correct the markup to match your real details (the drafted JSON-LD uses your config values).	Conflicting facts make assistants unsure which number or address is right, and a wrong number loses the call.	+4
Add a <label for=...> (or aria-label) to every field.	Labeled fields tell agents (and screen readers) what to type where. Placeholder text is not a label.	+4
Render these pages' content on the server or as static HTML.	Many AI crawlers read the raw HTML and do not run JavaScript, so script-built content can be invisible to them.	+2
Add geo, sameAs to the business markup.	These details answer common customer questions directly.	+2
Add 'Sitemap: <your site>/sitemap.xml' to robots.txt.	Crawlers find every page faster when robots.txt points to the sitemap.	+1.1
Keep it if spam is a problem, but make sure a phone number or email is always offered next to the form.	CAPTCHAs block spam bots but also block legitimate agents acting for a customer. Keep a phone or email path available.	+1

AI crawlers and your robots.txt

Crawler	Company	Purpose	Your site	Rule that applies
OAI-SearchBot	OpenAI	AI search / answers	Allowed	Allow: /
ChatGPT-User	OpenAI	AI search / answers	Allowed	-
Claude-SearchBot	Anthropic	AI search / answers	Allowed	-
Claude-User	Anthropic	AI search / answers	Allowed	-
PerplexityBot	Perplexity	AI search / answers	Blocked	Disallow: /
Perplexity-User	Perplexity	AI search / answers	Allowed	-
Googlebot	Google	Search engine (feeds AI answers)	Allowed	-
Bingbot	Microsoft	Search engine (feeds AI answers)	Allowed	-
GPTBot	OpenAI	Model training	Blocked	Disallow: /
ClaudeBot	Anthropic	Model training	Blocked	Disallow: /
Google-Extended	Google	Model training	Allowed	-
Applebot-Extended	Apple	Model training	Allowed	-
CCBot	Common Crawl	Model training	Allowed	-
meta-externalagent	Meta	Model training	Allowed	-
Bytespider	ByteDance	Model training	Allowed	-

Blocking AI search and user agents hides you from AI answers. Blocking training crawlers is a business choice and does not lower this score.

AI crawler access

Result	Check	What we found	Fix
FAIL	Let AI search and answer engines read your site	Blocked: PerplexityBot.	Remove the robots.txt rules that block PerplexityBot (see robots-ai-section.txt).
PASS	Let Google and Bing crawl your site	Googlebot and Bingbot may read your key pages.	-
PASS	robots.txt is present and readable	4 group(s)	-
WARN	robots.txt lists your sitemap	Sitemap line(s): 0	Add 'Sitemap: <your site>/sitemap.xml' to robots.txt.
NOTE	Model-training crawlers (your choice)	Blocked: GPTBot, ClaudeBot. Allowed: Google-Extended, Applebot-Extended, CCBot, meta-externalagent, Bytespider.	No change needed; this is your call.

Facts readable as text

Result	Check	What we found	Fix
PASS	Phone number as text and a tap-to-call link	(303) 555-0142 appears as text and as a tel: link.	-
PASS	Address as text	"123 Main St, Denver, CO 80204" appears as text.	-
FAIL	Business hours as text	No opening hours found as text. Hours appear to be shown as an image on /contact/; tools that read text cannot see them.	Write your hours as text (for example 'Mon-Fri 7am-6pm') on the contact page and footer.
PASS	Each service named in plain text	4 of 4 found.	-
PASS	Service area named in text	Mentions Denver metro.	-
PASS	Prices, price ranges or how pricing works	8 price mention(s), e.g. $79, $149, $1,600.	-
FAIL	Main content is in the HTML (not only added by JavaScript)	Little readable text without JavaScript on: /services/bathroom-remodels/.	Render these pages' content on the server or as static HTML.

Structured data

Result	Check	What we found	Fix
PASS	Valid JSON-LD structured data	13 item(s) found.	-
PASS	LocalBusiness (or Organization) markup	Plumber on / with name, address, phone and URL.	-
WARN	Hours, area, geo, profiles and price range in markup	Has: hours, areaServed, priceRange. Missing: geo, sameAs.	Add geo, sameAs to the business markup.
FAIL	Markup matches your real name, phone and address	markup phone (303) 555-0199 differs from your phone (303) 555-0142; the pages show (303) 555-0142 as text, not the markup number.	Correct the markup to match your real details (the drafted JSON-LD uses your config values).
PASS	Service or offer markup	2 Service/Offer item(s).	-
PASS	FAQ markup	FAQPage on /blog/signs-you-need-a-new-water-heater/.	-

Actionability for agents

Result	Check	What we found	Fix
PASS	A contact or booking page linked with a plain link from the home page	Linked: /contact/.	-
FAIL	Form fields have labels	4 of 4 field(s) have no label (placeholder text only) on /contact/: name, email, phone, message.	Add a <label for=...> (or aria-label) to every field.
PASS	Forms submit with a normal action and button	Forms have an action, a method and a submit button.	-
PASS	Tap-to-call or email links	tel: link found.	-
WARN	CAPTCHA on forms (trade-off)	CAPTCHA on /contact/.	Keep it if spam is a problem, but make sure a phone number or email is always offered next to the form.
NOTE	Agent / MCP descriptor files (emerging, optional)	None found. That is normal today.	Optional. Revisit when your booking software offers one.

Basics

Result	Check	What we found	Fix
PASS	Secure (https) website	https://www.example.com uses https (checked your configured website address).	-
PASS	XML sitemap	11 URL(s) in 3 file(s).	-
PASS	Server response time	Median time to first byte 2 ms over 12 page(s) (measured from where this check ran).	-

llms.txt

Result	Check	What we found	Fix
FAIL	Publish an llms.txt file	/llms.txt returns 404.	Upload the drafted llms.txt to the root of your site.
NOTE	llms-full.txt (optional)	Not found (optional).	Optional.

Files drafted for you

llms.txt: a starting llms.txt built from your page titles and descriptions. Review, then upload to https://www.example.com/llms.txt.
localbusiness-jsonld.html: LocalBusiness structured data built only from the facts in your config. Paste it into your home page <head>.
robots-ai-section.txt: a suggested robots.txt section that lets AI search and answer engines in; training crawlers are left as your choice.
Saved in: reports/agent_ready_auditor/fixes-2026-10-15

About this audit

Agent readiness is an emerging area. AI companies change their crawlers and how they use websites often, and standards such as llms.txt and agent descriptor files are proposals that not every AI tool reads. These checks are current best practices that make your business easier for assistants to find, understand and contact. They do not guarantee that any AI tool will recommend you, or any ranking or traffic.

Generated Oct 15, 2026 01:06 AM MDT by Agent Army Suite. Drafts wait for your approval; nothing was sent automatically unless your policy allows it.