Structured data
JSON-LD, Schema.org types, Speakable markup and sameAs references.

Products / Glippy
Glippy runs 240+ checks in 16 categories and gives a page a GEO score from 0 to 100 with a letter grade. It looks at what LLM crawlers, AI agents and interfaces like ChatGPT, Perplexity and Google AI Overviews need from a page.
0–100
A score on its own does not fix anything. Glippy breaks it down per category, shows every individual check, and puts the weakest areas first, so the first fix is also the one that moves the score most.
GEO score
72
Grade
B
Fixes queued
9
illustrative data
16 categories, 240+ checks
From structured data and semantic HTML to how many citable facts a page carries per paragraph. Each category is scored on its own.
JSON-LD, Schema.org types, Speakable markup and sameAs references.
Heading hierarchy, semantic elements and the content-to-markup ratio.
Lang attributes, alt text, ARIA landmarks, link text and form labels.
Link density, navigation structure, breadcrumbs and anchor links.
Title, description, canonical URL, Open Graph and language tags.
Server-side rendering, JavaScript frameworks, blocked AI crawlers and robots meta tags.
Author information, publication dates, organisation signals and legal pages.
FAQ content, tables, lists, lead paragraphs and answer-first sections.
Image dimensions, lazy loading, render-blocking scripts and DOM size.
WebMCP tools, their descriptions, and whether forms are ready for agents.
Differentiation, proof points and the value propositions LLMs repeat when they compare.
Publication and update dates, and stale dates an answer engine would distrust.
Substance against filler, and claims paired with evidence.
Sources, attribution for statistics and checkable product facts.
Depth for the page type, subtopic coverage and defined key terms.
Alt text quality, figure captions, video transcripts and media schema.
Three ways to run it
The same checks everywhere. Pick the one that fits how you work.
Your own pages, competitors or a client site before the work starts. Scan history and multi-page scans. It runs in your browser and sends no page content to my servers.
An overall score, 16 scored categories, AI-crawler rules and a prioritised list of fixes, right in the chat. Compare up to eight sites, or ask ChatGPT to draft the fixes.
11 tools for domains, URLs, sitemaps up to 50,000 URLs per run, robots.txt, llms.txt and live crawler access. Works in Claude, Cursor, VS Code, Windsurf, Zed, Cline and Warp.
What it checks, what it costs, and what it does not do.
A number from 0 to 100 for how well a page is prepared for generative engines: whether AI crawlers may fetch it, whether the content survives without JavaScript, and whether the facts on it are easy to extract and cite. It is a readiness check, not a ranking prediction.
The ChatGPT plugin is free and needs no account. The Chrome extension has a free tier with 11 scans and a one-time Pro licence. The MCP server comes with 11 free trial runs, then a one-time lifetime licence.
Yes, the MCP server has a tool that reports whether one exists and what it contains. Whether the file helps is another question: I tested llms.txt on more than 150 domains and saw no LLM bot use it. The test is on the blog.
Or install the extension and score the next ten pages you open, including your competitors'.