Agent readiness glossary

1.A2A (Agent2Agent protocol)

An open protocol, maintained by the Linux Foundation, for agents to discover each other through Agent Cards and exchange tasks.See: /standards/a2aSource: A2A specification

2.Accountable crawler

A designation Cloudflare introduced on 15 September 2026 for crawler operators that meet stated opt-out and reporting requirements. It works with Cloudflare's setting that disallows AI training while keeping search access.See: /university/setting-an-ai-crawler-policySource: Cloudflare blog, Accountable mixed-use AI crawlers

3.Agent Card

A JSON document in which an A2A agent describes its identity, capabilities and endpoints, published at a well-known URI.See: /standards/a2aSource: A2A specification

4.Agent experience (AX)

How it goes for an AI agent that tries to use a site or product, the agent equivalent of user experience.

5.Agent journey

The sequence of steps an agent takes to complete a task on a site, such as finding a price and reaching a sign-up form. ora.ai Journey records these.See: /tools/ora-ai

6.Agent readability

Whether an agent can read and correctly interpret a site's content, for example because pages are server-rendered, text-based and available as markdown.

7.Agent readiness

How well a site or product can be discovered, read, used and paid by AI agents acting for people.See: /

8.Agent skill

Reusable instructions that give an AI agent the context for a particular job, often a plain-language file. Vercel's September 2026 report on the skills.sh registry uses this definition.See: /newsSource: Vercel, State of agent skills

9.Agent-facing site

A version of a website served to agents rather than people, with the same facts in a form agents parse more easily.

10.Agentic AI Foundation (AAIF)

A Linux Foundation-directed body that governs open agent infrastructure projects. A2A joined it in August 2026; the A2A announcement names MCP, goose and AGENTS.md as sibling projects.See: /standards/a2aSource: A2A project blog

11.Agentic commerce protocols

Protocols for agent-initiated purchases. ora.ai's AgentReady specification and Cloudflare's scanner list ACP, UCP and MPP alongside x402. This edition does not cover them in depth.See: /standards/x402

12.AgentReady

An MIT-licensed open specification published by ora.ai at agentready.org that names what to implement for agent readiness in five categories, with stable requirement IDs such as AR-DISC-01. v1.0.0 was published on 24 April 2026; ora.ai describes it as a specification, not a scoring system.See: /tools/ora-aiSource: ora.ai, AgentReady announcement

13.agents.json

An open specification from Wildcard AI that adds flows and links on top of OpenAPI, published at /.well-known/agents.json.See: /standards/agents-jsonSource: agents.json repository

14.AEO and GEO

Answer engine optimisation and generative engine optimisation: work aimed at appearing in AI-generated answers. Different from agent readiness.See: /notes/agent-readiness-and-aeo

15.AI crawler

An automated client operated by an AI company that fetches web pages for training, search or live answers, identified by a user agent name such as GPTBot or ClaudeBot.See: /standards/robots-txt-ai-crawlers

16.Client ID Metadata Document

An OAuth approach in which a client identifies itself with a URL that points to a metadata document it publishes. The MCP specification of 28 July 2026 adopts it and deprecates Dynamic Client Registration.See: /dispatches/mcp-2026-07-28-specification-explainedSource: MCP blog, The 2026-07-28 specification

17.Content signals

Declarations of how content may be used by AI, such as Cloudflare's content-signal: ai-train=yes, search=yes, ai-input=yes response header.See: /tools/cloudflareSource: Cloudflare Markdown for Agents docs

18.Declarative WebMCP

The WebMCP form in which tools are synthesised from existing HTML forms.See: /standards/webmcpSource: WebMCP repository

19.Elicitation

An MCP feature in which a server asks the user, through the client, for more information during an operation. Under the 2026-07-28 specification the server returns an input-required result and the client retries, instead of holding a stream open.See: /standards/mcpSource: MCP specification

20.File check

A test that requests one URL and inspects the response, such as confirming that /robots.txt returns HTTP 200 or that a page returns markdown when asked. Contrast with a task run.See: /university/file-checks-and-task-runs

21.Grade band

A named range of scores. ora.ai's methodology publishes six bands, from A+ (95–100, Leading) to F (0–27, Unusable).See: /university/how-to-read-a-readiness-scoreSource: ora.ai methodology

22.HTTP 402 Payment Required

An HTTP status code that x402 uses to tell a client that payment is needed before retrying.See: /standards/x402Source: x402.org

23.HTTP Message Signatures

The IETF method (RFC 9421) for signing parts of an HTTP message; the basis of Web Bot Auth.See: /standards/web-bot-authSource: RFC 9421

24.Imperative WebMCP

The WebMCP form in which JavaScript registers tools with a name, description, input schema and function.See: /standards/webmcpSource: WebMCP repository

25.JSON-LD

A JSON format for structured data, commonly with Schema.org vocabulary, embedded in pages so machines can read facts such as products and offers.

26.JSON-RPC 2.0

A lightweight remote procedure call format in JSON. MCP messages use it, and it is one of the three transport bindings in the A2A specification.See: /standards/mcpSource: MCP specification

27.Key directory

A file at /.well-known/http-message-signatures-directory where a bot operator publishes the public keys used to verify its signed requests.See: /standards/web-bot-authSource: Cloudflare Web Bot Auth docs

28.llms.txt

A markdown file at the site root that summarises a site for language models and links to key pages. Version 2 of the proposal (August 2026) adds link relations so pages can point to their markdown version and to the llms.txt that covers them.See: /standards/llms-txtSource: llmstxt.org

29.Markdown negotiation

Serving a markdown version of a page when a client sends Accept: text/markdown.See: /tools/cloudflareSource: Cloudflare Markdown for Agents docs

30.MCP (Model Context Protocol)

An open protocol that connects LLM applications to servers offering tools, resources and prompts over JSON-RPC 2.0.See: /standards/mcpSource: MCP specification

31.Multi round-trip request (MRTR)

The pattern in the MCP 2026-07-28 specification where a server that needs more input returns a result saying so, and the client retries the request with that information.See: /dispatches/mcp-2026-07-28-specification-explainedSource: MCP blog, The 2026-07-28 specification

32.OAuth 2.0

The authorization framework used to give an application or agent scoped, revocable access to a user's account without sharing a password. Readiness checklists place it in the trust layer.See: /checklist#l5

33.OpenAPI

A machine-readable description format for HTTP APIs. agents.json builds on it by adding flows and links, and readiness checklists list it under capabilities.See: /standards/agents-jsonSource: agents.json repository

34.Pay per crawl

A Cloudflare feature, in private beta, that lets AI crawlers pay to access content.See: /tools/cloudflareSource: Cloudflare Pay per crawl docs

35.Readiness scanner

A tool that takes a public URL and reports which agent readiness signals a site exposes. We score two like-for-like scanners, ora.ai Scan and Cloudflare Is It Agent Ready.See: /calculator

36.Readiness score

A number summarising which agent readiness signals a scanner found on a site. It does not measure AI search visibility or conversion.See: /method

37.robots.txt

A file at /robots.txt that asks crawlers to follow path rules per user agent; standardised as RFC 9309 and not a form of access authorization.See: /standards/robots-txt-ai-crawlersSource: RFC 9309

38.Server-side rendering (SSR)

Producing a page's full HTML on the server, or at build time as prerendering, so a client that does not run JavaScript still receives the content.See: /university/how-agents-read-a-page

39.Streamable HTTP

The HTTP transport Cloudflare documents for remote MCP servers, with OAuth for authorization.See: /tools/cloudflareSource: Cloudflare MCP docs

40.Task run

A test in which an agent is given an intent, such as finding a price, and its path through the site is recorded, including where it stalled. ora.ai Journey records task runs.See: /university/file-checks-and-task-runs

41.Verified bot

A bot whose identity a CDN or platform has confirmed, for example through Web Bot Auth signatures, so sites can treat it differently from unknown automation.See: /tools/vercel

42.Web Bot Auth

A method for bots to sign HTTP requests so sites can verify the operator, defined in IETF Internet-Drafts.See: /standards/web-bot-authSource: Cloudflare Web Bot Auth docs

43.WebMCP

A W3C Community Group proposal that lets a web page register tools an in-browser agent can call.See: /standards/webmcpSource: WebMCP repository

44.Well-known URI

A reserved path under /.well-known/ where a client can find a file at a predictable location, such as /.well-known/agents.json or a Web Bot Auth key directory.See: /standards/agents-json

45.x402

An open standard that uses HTTP 402 so a client can pay per request, typically with stablecoins, and retry.See: /standards/x402Source: x402.org

46.XML sitemap

A file that lists a site's URLs for crawlers, usually referenced from robots.txt. llms.txt is a short curated index and does not replace it.See: /checklist#l1

Source: sources as on each linked page · Reviewed Sep 2026