Agent view — served from /glossary.md.
Reference
AI visibility glossary
Twenty-seven terms used across AI visibility and agent readiness, defined in one or two sentences each. Where a term is disputed or its value unproven, the definition says so rather than presenting a contested idea as settled fact.
TermDefinition
Agent readiness
How well a website works for AI agents: whether they can discover it, parse its content, and complete actions on it. Measured separately from SEO and page speed.
AEO
Answer Engine Optimization. Structuring content so AI systems can extract a direct answer from it and cite the source.
GEO
Generative Engine Optimization. In practice the same discipline as AEO; no agreed definition separates them.
llms.txt
A plain-markdown file at a site root giving AI models a curated map of its most important pages. Its efficacy is disputed.
llms-full.txt
A companion file concatenating the full markdown of key pages into one document a model can ingest in a single request.
agents.json
A machine-readable declaration of a site’s identity, capabilities, and the actions agents may take.
Content negotiation
Serving a different representation of a page based on a request header, typically markdown instead of HTML when an AI asks for it.
Accept: text/markdown
The HTTP request header an AI client sends to ask for a markdown version of a page.
Content-Signal
A robots.txt directive declaring how content may be used: for training, for search, or as input to an AI answer.
GPTBot
OpenAI’s training crawler. Distinct from ChatGPT-User, which fetches pages to cite in live answers.
ChatGPT-User
OpenAI’s citation agent, which fetches a page because a user asked a question. Blocking it removes you from live answers.
Schema drift
When structured data no longer matches the visible page, such as a JSON-LD price differing from the displayed price.
WebMCP
An emerging browser standard letting a site expose its functions as typed tools an AI agent can call. Currently a Chrome origin trial.
MCP
Model Context Protocol. A standard for connecting AI assistants to external tools and data sources.
MCP server card
A file at /.well-known/mcp.json advertising that a site runs an MCP server agents can connect to.
Agentic browsing
An experimental Lighthouse audit category, added in version 13.3, checking whether a page is usable by AI agents.
Accessibility tree
The semantic structure assistive technology and AI agents navigate, as opposed to the visual layout a person sees.
BLUF
Bottom Line Up Front. Answering the question in the opening of a passage rather than after context and background.
Atomic content
Content written so each section stands alone and retains its meaning when extracted without its neighbours.
Query fan-out
When an AI breaks a single user prompt into multiple sub-questions and retrieves sources for each.
How often AI engines cite a given domain for a defined set of questions. The primary AI visibility metric.
AI Overviews
Google’s generated answers at the top of search results. No special structured data is required to appear in them.
Web Bot Auth
A mechanism letting a legitimate bot cryptographically sign its requests to prove identity, rather than relying on IP allow-lists.
x402
A proposal reviving the HTTP 402 Payment Required status so agents can pay for resources programmatically.
Rubric version
The pinned version of a scoring definition. Re-audits reuse the original so before-and-after comparisons stay valid.
Tier 1 check
A deterministic check that produces the same result on every run and contributes to the headline score.
Tier 2 check
An LLM-assisted judgement reported as an advisory flag with evidence. Carries no points.
See how AI reads your site
Eight checks against your domain in about fifteen seconds. No email.