What the scan checks

Each item below is a check that the free InspectGeo scan runs on your site. Open one to see what it looks at and why it matters for AI to read, cite and act on your site.

Citable by AI

• AI search crawlers allowed

  ChatGPT, Perplexity, Claude, Google and Bing only cite what their search crawlers can read. If robots.txt or a firewall blocks those crawlers, the sit...

• Indexable page

  AI answers with web search rely on the Google and Bing indexes. A home page marked noindex (often left over from when the site was under construction,...

• Content readable without JavaScript

  AI crawlers read the HTML the server delivers, and most of them do not run JavaScript. A site built as a single-page application (React, Vue and many...

• Site identity

  To cite a site, an AI needs to know who is behind it; Google calls this E-E-A-T (experience, expertise, authoritativeness and trust). The signals a cr...

• Title, H1 and description

  The page title, the H1 and the meta description are the summary that search engines and AIs read first to learn what the page is about. Without them,...

• Freshness signal

  Search engines and AIs prefer sources that show they are up to date, especially on topics that change over time. The update date in structured data (d...

Discoverability

• robots.txt

  Without robots.txt an agent has nowhere to read the rules of the site or find the sitemap. It is the most basic foundation of discovery: everything el...

• XML sitemap

  Without a sitemap, an agent only discovers the site's pages by following links one by one — slow, incomplete and dependent on how the site is linked i...

• Link header

  The HTTP Link header exposes typed relations without requiring the agent to download and parse the whole HTML — it can find the API catalog, the sitem...

• DNS-AID (DNS-based discovery)

  It is the lowest possible discovery layer: it happens at DNS resolution, before any HTTP request. An agent or discovery service can ask DNS 'does this...

Content

• llms.txt

  llms.txt is today the most adopted way for a site to introduce itself to an agent: a plain-text summary, with the addresses of what matters, without t...

• Markdown negotiation

  HTML carries a lot of noise (navigation, scripts, styles) that does not help a language model understand the content — it only raises the token cost a...

• Structured data without conflicts

  Agents and search engines read the structured data (JSON-LD) of a page to learn who the site is: name, address, logo. When two blocks say that in diff...

Bot Access Control

• AI bot rules

  Generic bots and AI bots have different interests — one indexes for search, the other may be collecting training data or fetching context to answer a...

• Content Signals

  Blocking or allowing a bot in a traditional robots.txt is an all-or-nothing decision about access. Content-Signal is more granular: it lets you say, f...

• Web Bot Auth

  User-Agent is just a string any bot can forge — it proves nothing. Web Bot Auth uses HTTP cryptographic signatures to actually prove that a request co...

API, Auth, MCP & Skills

• API Catalog

  An agent that wants to interact with your site should not have to guess endpoints or dig through an HTML developer page. The API Catalog is the sitema...

• OAuth/OIDC discovery

  OAuth/OIDC discovery lets an agent automatically find out where and how to authenticate (authorization and token endpoints) without manual documentati...

• OAuth Protected Resource

  It complements OAuth discovery by stating specifically which authorization server protects which resource and which scopes are accepted. Without this...

• auth.md

  When a site has authenticated areas, an agent that wants to operate on the user's behalf needs to know how to register and obtain credentials — withou...

• MCP Server Card

  MCP has become the de facto standard that production agent products use to connect to external tools and data. To be transparent: MCP's official disco...

• A2A Agent Card

  A2A solves a different problem from MCP: it is not an agent calling a tool and getting an immediate answer, it is an autonomous agent delegating a who...

• Web Skills Index

  Agent Skills (agentskills.io) is the open skill format: a folder with a SKILL.md that an agent loads when a task calls for it. For the web, discovery...

• WebMCP

  It is the only protocol designed for agents that browse the web with a real browser instead of calling server APIs. Exposing document.modelContext let...

E-commerce

• x402 (HTTP payment)

  x402 reuses the HTTP 402 status code (Payment Required), which has been in the standard forever but never really used, to let an agent pay for a resou...

• MPP (Machine Payment Protocol)

  Unlike x402 (a request-response flow), MPP declares payment information directly in the API's OpenAPI contract — by reading the specification the agen...

• UCP

  UCP aims to be a common commerce profile that different agents and platforms understand without a specific integration — the goal is that a generic sh...

• ACP (Agentic Commerce)

  ACP defines how a shopping agent carries out an entire transaction on the user's behalf — product discovery, checkout and payment — end to end, inside...