Glossary
In short
40 terms used on this site, each defined in a sentence or two and followed by a sourced, dated example. They run from initial HTML and plain fetch through soft 404, llms.txt, markdown negotiation and JSON-LD to answer provenance and token reduction. Each term has its own page.
Crawl edition 2026-11-04, page updated 2026-11-04. How the viewer, the index and the scores work.
- Accept header
- The request header that lists media types a client will take, with optional q weights.
- AFDocs
- The Agent-Friendly Documentation spec and its open-source checker, used by Mintlify, ReadMe and GitBook docs scores.
- Agent 404
- A 404 response written for agents, often markdown, that links the index and similar pages so the agent can recover.
- Agent score (docs)
- A docs-specific readiness score computed from AFDocs-style checks, published by docs vendors.
- Agent-optimised mirror
- A separate version of pages served only to AI user agents at the same URLs.
- agents.md
- On Shopify, the canonical agent-discovery document of a store, mirrored at /llms.txt and /llms-full.txt.
- Answer provenance
- Which fetched page an agent answer can be traced to.
- App shell
- A near-empty HTML document whose content is injected by JavaScript after load, typically a single div with id root or app.
- Content negotiation
- The HTTP mechanism where the client states preferred formats in the Accept header and the server picks a representation for the same URL.
- Content-Signal
- A declaration of permitted AI uses (ai-train, search, ai-input), sent as a robots.txt line or a response header.
- Edge conversion
- Turning origin HTML into another format at the CDN, without changing the origin.
- Fenced code block
- Code set off by triple backticks with a language tag, so agents can separate code from prose.
- File list
- An H2 section in llms.txt holding a markdown list of links in the form [name](url), optionally followed by a colon and notes.
- HTTP Link header
- A response header carrying link relations (RFC 8288), so a CDN can advertise files without changing markup.
- Hydration
- The step where client JavaScript attaches event handlers to server-rendered HTML. The content is already present before it happens.
- Initial HTML
- The HTML document a server returns for a URL before any JavaScript runs. A plain-fetch agent reads only this, so the answer has to be in it.
- JSON-LD
- Structured data written as JSON in a script tag of type application/ld+json, using schema.org types.
- Lighthouse llms.txt audit
- A Chrome Lighthouse check in the Agentic Browsing category that flags a server error on /llms.txt and marks a 404 as not applicable.
- Linked arrival
- A fetch of a machine file that followed a link rather than a guessed path.
- llms-full.txt
- A companion file, not defined in the llmstxt.org text, that concatenates the full markdown of a site in one document.
- llms.txt
- A markdown file at /llms.txt that names a site, summarises it and lists links to LLM-friendly pages. Only the H1 is required.
- Markdown negotiation
- Content negotiation where Accept: text/markdown returns the page as markdown with Content-Type text/markdown.
- Markdown twin
- A markdown copy of a page at a predictable URL, such as page.md or page.html.md.
- OpenAPI
- A machine-readable description of an HTTP API, usually openapi.json or openapi.yaml.
- Optional section
- An H2 named Optional in llms.txt that holds secondary links. v2 keeps it as a convention and drops its special meaning for tools.
- Plain fetch
- An HTTP GET that reads the response body without executing scripts, the way most agent web tools and curl read pages.
- Prerendering
- Rendering pages to static HTML at build time so every request gets finished content.
- Real status code
- An HTTP status that matches what happened: 200 for a page, 404 for a missing path, 429 for throttling. AgentReady makes this a MUST.
- rel="alternate" type="text/markdown"
- A link relation in HTML or an HTTP Link header that points from a page to its markdown version. Added to llms.txt v2.
- rel="describedby"
- A link relation that points from a page to the llms.txt file covering it. Added in llms.txt v2.
- Rendered HTML
- The DOM after a browser has run the page scripts. A human and a JavaScript-running agent see it; a plain-fetch client does not.
- schema.org
- The shared vocabulary of types and properties (Organization, Product, TechArticle) used in JSON-LD.
- Server Components
- React components that render on the server; the default for pages and layouts in the Next.js App Router.
- Server-side rendering (SSR)
- Rendering a page to HTML on the server per request, so the response carries the content.
- Soft 404
- A missing page that returns 200 with an error message in the body. Agents and crawlers cannot tell it from a real page.
- Subpath llms.txt
- An llms.txt below the root that covers the pages under its path; v2 says the most specific file applies.
- Token reduction
- How many fewer model tokens a page costs as markdown than as HTML.
- Universal rendering
- Nuxt term for rendering on the server first and hydrating in the browser.
- Vary: Accept
- A response header telling caches that the body depends on the Accept header, so HTML and markdown are stored separately.
- x-markdown-tokens
- A response header Cloudflare adds with the estimated token count of the markdown body.
Sources
- llms.txt proposal (llmstxt.org) https://llmstxt.org/, read 4 November 2026
Last reviewed 4 November 2026