Agent Readiness on seoboosterpro.com (Honest Dogfood)

Quick answer: We improved agent discovery on seoboosterpro.com with Content Signals, HTTP Link headers, an RFC 9727 api-catalog, real Agent Skills, and Markdown negotiation. We did not fake MCP, OAuth, or payment protocols we do not operate.

Cloudflare’s Agent Readiness scanner grades how well a site talks to AI agents. Our first pass looked weak on Link headers, Content Signals, Markdown negotiation, and protocol discovery, even though we already shipped strong AI discovery assets: curated /llms.txt, /entitymap.json, and robots lines for both.

This post is the dogfood write-up: what we added, what we refused to invent, and how that maps to SEO Booster on WordPress.

What scanners actually measure

At a high level, Agent Readiness asks three questions:

May agents use this content, and how? (Content Signals in robots.txt)
Where are the machine-readable endpoints? (HTTP Link headers, not only HTML <link> tags)
What protocols and catalogs exist? (.well-known discovery such as api-catalog and Agent Skills)

Commerce and OAuth checks matter when you run agent-callable APIs or payments. A product marketing site usually should not invent those.

What we already had

Before this work, seoboosterpro.com already published:

/llms.txt (and a Danish twin)
EntityMap at /entitymap.json and /entitymap.html
LLMS: and EntityMap: lines in robots.txt
HTML head discovery links for llms.txt and Entity Map

Those help real systems today. Cloudflare’s scanner still wanted newer signals: Content Signals, response Link headers, Markdown negotiation, and structured .well-known catalogs.

What we added

Content Signals in robots.txt:

Content-Signal: search=yes, ai-train=no, ai-input=yes

Search indexing is welcome. Training on our site content is not. Inference and citation with attribution are OK.

HTTP Link headers (nginx) advertising:

/llms.txt
/entitymap.json
/.well-known/api-catalog
/.well-known/agent-skills/index.json
/.well-known/agent-knowledge.json
/index.md as text/markdown

Discovery catalog at /.well-known/api-catalog: an RFC 9727-style linkset over assets we actually serve (Entity Map, docs, llms.txt, skills, knowledge card). No fake OpenAPI for a product API we do not host on this domain.

Agent Skills (discovery RFC v0.2.0) at /.well-known/agent-skills/index.json, each with a SKILL.md and sha256: digest:

site-discovery — how to use our discovery files (and that MCP is absent)
wordpress-ai-readiness — checklist pointing at our guides and plugin docs
human-support — route hard cases to /support/ (no agent ticket API)

Honest knowledge card at /.well-known/agent-knowledge.json: publisher, content policy, discovery URLs, and protocols.mcp: null.

Markdown: site brief at /index.md, plus .md mirrors for the AI guide cluster (for example /seo/ai-search-readiness-wordpress.md). nginx serves markdown when Accept includes text/markdown.

What we deliberately do not ship

Protocol / check	Why we skip it
MCP server card	We do not run MCP on this marketing site
OAuth / OIDC / Auth.md	No agent-auth APIs here
A2A agent card	No agent-to-agent runtime
DNS-AID	High operational cost; robots + Link + well-known cover the practical path
WebMCP / commerce protocols	Not applicable to Freemius checkout on a static site

Passing a scan by lying about endpoints teaches agents the wrong model of your product. We would rather leave those checks failing.

Two layers: marketing site vs WordPress plugin

Layer	Job
seoboosterpro.com	Honest discovery for our own brand, docs, and Entity Map
SEO Booster (WordPress)	Help your site publish llms.txt, Pro Entity Map, optional Markdown URLs, and Link discovery signals

If you run WordPress, start with the AI search readiness hub, then llms.txt and Entity Map vs other indexes. Plugin docs: llms.txt tool and Entity Map.

How to verify

curl -sI https://seoboosterpro.com/ | grep -i '^link:'
curl -sI -H 'Accept: text/markdown' https://seoboosterpro.com/ | grep -i content-type
curl -s https://seoboosterpro.com/.well-known/api-catalog | head
curl -s https://seoboosterpro.com/.well-known/agent-skills/index.json | head
curl -s https://seoboosterpro.com/robots.txt | grep Content-Signal

Then re-run radar.cloudflare.com/scan/agents. Expect MCP and OAuth-style checks to remain failed unless you later ship those systems for real.

Frequently asked questions

What is Agent Readiness in this context?

Agent Readiness is a practical checklist used by scanners such as Cloudflare Radar for Agents. It looks for signals that declare content use preferences, advertise machine-readable endpoints (for example HTTP Link headers), and publish discovery files under /.well-known/. It is a preparedness score, not a Google ranking factor.

Did you publish an MCP server to pass the scan?

No. We do not run an MCP server on seoboosterpro.com, so we do not publish an MCP server card. Inventing one would mislead agents. We publish an honest agent-knowledge.json that states MCP is unavailable, plus api-catalog and Agent Skills that point at real assets.

How does this relate to SEO Booster for WordPress?

This article is dogfood on our marketing site. On customer WordPress sites, SEO Booster already helps with curated llms.txt, Pro Entity Map, optional Markdown URLs, and HTTP Link discovery signals. Marketing-site Agent Readiness and plugin Tools are complementary layers, not the same feature set.

Will Content Signals or Link headers guarantee AI citations?

No. Clear discovery files and policies make it easier for agents to find and correctly use your content. They do not guarantee that any specific AI product will crawl, cite, or prefer your pages.