aeo-geo-site
Skill8 starsChanged 3 days agoMITView on GitHub
use this when building or updating a website for SEO plus AEO/GEO (AI assistants, ChatGPT, Copilot, Gemini) so pages are machine-readable, intent-clear, and citable
In this repository28 other files it gives its agents.
- whetstone-signalSkill
- design-a-grok-botSkill
- make-bot-uiSkill
- overheard-setupSkill
- overheardSkill
- routine-healthcheckSkill
- transcript-healthcheckSkill
- design-taste-frontendSkill
- diagram-designSkill
- motion-pickerSkill
- redesign-existing-projectsSkill
- engineering-playbookSkill
- show-meSkill
- chatgpt-images-browser-playbookSkill
- ai-citation-sourcesSkill
- ai-search-gap-loopSkill
- exact-match-intent-seoSkill
- markdown-twinSkill
- shorts-reframeSkill
- coachSkill
- draftSkill
- no-ai-slopSkill
- outlineSkill
- social-copywritingSkill
- source-checkSkill
- top-editSkill
- x-articleSkill
- youtube-niche-reportSkill
The file
We split this skill into the rules it gives an agent
8 rules. About 0 are common, 7 are distinctive, and 1 is very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈7
Distinctive · estimate
This author's own choices. Worth a look.
1
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 12
Use this whenever you ship or restyle site pages meant to be found, understood, cited, or recommended by AI search, assistants, browsers, …
- Very specific: line 30
Never cloak: same HTML to bots and users.
- Distinctive: line 46
After meaningful ships, rescan agent-readiness (e.g.
- Distinctive: line 55
Use-case framing (“ready in 20 minutes”, “works with your existing CI”) over vague slogans.
- Distinctive: line 58
Prefer one clear answer per section.
- Distinctive: line 74
No exaggerated claims, no fake social proof, no Ivy-league wallpaper, no “#1” without a source.
- Distinctive: line 101
Read brand / design law and existing schema patterns on the site.
- Distinctive: line 105
Verify with a live fetch (HTML + markdown Accept) and an agent-readiness scan after merge.
≈ “Common” is an estimate: at least 1 in 5 of the 364,980 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
AEO / GEO site (Microsoft + execution)
Source frame: Microsoft Advertising, Jennifer Myers (author), and Paul Longo (executive sponsor), From discovery to influence: A guide to AEO and GEO (Jan 2026), as summarized in Alex Groberman. Primary PDF: https://about.ads.microsoft.com/content/dam/sites/msa-about/global/common/content-lib/pdf/from-discovery-to-influence-a-guide-to-aeo-and-geo.pdf
Use this whenever you ship or restyle site pages meant to be found, understood, cited, or recommended by AI search, assistants, browsers, or agents. SEO remains the foundation. AEO adds clarity. GEO adds trust.
Shift to optimize for
| Layer | Goal | Old SEO bias | New bar |
| SEO | Be found | Keywords, rankings, clicks | Crawlable, fast, indexable pages |
| AEO | Be understood | Thin keyword copy | Clear, structured, extractable answers |
| GEO | Be chosen / cited | Soft claims | Credible, consistent, justified facts |
Competition moves from “being found” to “being chosen.” Visibility is earned by how clearly AI systems understand the entity, trust the brand, and can act on the data.
Three data planes (control all that apply)
- Crawled web data — what AI learned and what live search finds: brand reputation, category authority, expert mentions, crawlable HTML.
- Structured feeds / APIs — catalogs, sitemaps,
llms.txt, markdown twins, JSON-LD, public APIs. Precision and control live here. (Retail: product feeds. Content sites: entity pages, video objects, FAQ, design law, markdown negotiation.) - Live website data — what an agent sees on visit: current copy, media, reviews, working CTAs. If the live page lies or fails, the agent fails even if feeds were perfect.
Never cloak: same HTML to bots and users. Extra formats (markdown via Accept, llms.txt) are additive, not different facts.
Three action pillars
1. Technical foundations and structured data
AI needs structure and consistency more than clever prose.
- Server-rendered HTML first. Honor
robots, sitemap, canonicals,dateModifiedwhen content changes. - JSON-LD that matches visible content. Prefer real schema types for the product:
- Content / media sites:
Organization,WebSite,WebPage,FAQPage,VideoObject,ImageObject,BreadcrumbList,ContactPoint - Retail (when applicable):
Product,Offer,AggregateRating,Review,Brand,ItemList
- Content / media sites:
- Machine-readable twins where useful: markdown negotiation,
llms.txt, publicdesign.mdor equivalent brand law for agents. - Dynamic fields stay fresh: price/availability/SKU/GTIN for commerce; titles, publish dates, canonical URLs, AI-use disclosures for content.
- Localized signals when relevant:
inLanguage,priceCurrency, address that matches real ops. - Images: descriptive alt text;
ImageObjectwhen it adds citability. - After meaningful ships, rescan agent-readiness (e.g. is-agentic) and fix Essential misses before soft polish.
2. Intent-driven content enrichment (AEO)
Optimize for questions and extraction, not keyword stuffing.
- Front-load: who it is for, what problem it solves, what it is (and is not).
- Headings that mirror real reader / buyer / user questions.
- Modular, citable blocks: short paragraphs, Q&A, feature lists, comparison only when honest and on-brand.
- Use-case framing (“ready in 20 minutes”, “works with your existing CI”) over vague slogans.
- Video / media pages: transcript or plain-language summary, still + outbound watch URL, VideoObject.
- Relationships when real: “goes with”, related recipes or docs pages, related FAQs. Do not invent entities.
- Prefer one clear answer per section. Avoid em-dashes, hedge piles, and low-trust hype.
SEO vs AEO vs GEO in one line:
- SEO: “waterproof rain jacket”
- AEO: “lightweight packable waterproof rain jacket with ventilation”
- GEO: “best-rated by X, 4.8 stars, 180-day returns” (only if true and verifiable)
3. Trust and credibility (GEO)
AI systems prioritize verifiable truth and penalize low-trust language.
- Consistent entity: same name, same facts, same URLs across site, social profiles, video channels, schema.
- Proof you can show: reviews, press, certifications, expert mentions, official sameAs links — only real ones.
- FAQ and help content that answers hard questions honestly (including “no”, “not yet”, disclosures).
- Contact / trust pages when recommended (email, privacy, no fake forms).
- No exaggerated claims, no fake social proof, no Ivy-league wallpaper, no “#1” without a source.
- Brand voice stays calm and specific. Copy locks and design law beat trend copy.
Pre-ship checklist (every page or wave)
- Primary reader job is named (developer evaluating, agent citing, buyer comparing, …)
- H1 and first viewport answer that job without fluff
- Visible facts match JSON-LD and any markdown twin
- FAQ or Q-shaped headings where intent is question-shaped
- Images have useful alt; no type burned into photos unless brand allows it
- CTA points at the real product surface (signup, docs, buy URL, contact) — not a maze
noindexonly for experiments; production pages in sitemap when indexable- No cloaking; Accept negotiation serves the same facts
- Trust lines are true and on-brand (disclosures, AI-use notes, location, team)
- Soft claims and competitor digs stay off
Anti-patterns
- Ranking-only pages with nothing an assistant can quote
- Schema that invents ratings, products, or team members
- Different bot HTML vs user HTML
- Paywall / survey funnels as “content”
- Keyword stuffing, unverifiable superlatives, fear-based or comparison claims when brand forbids them
- Shipping a feed or schema type you cannot keep fresh
How to run on a task
- Read brand / design law and existing schema patterns on the site.
- Name the intent cluster and the entity (recipe, docs page, org, product).
- Draft or restyle with pillar 2 copy shape first, then pillar 1 markup, then pillar 3 trust links.
- Wire markdown /
llms.txt/ sitemap as needed. - Verify with a live fetch (HTML + markdown Accept) and an agent-readiness scan after merge.
Retail-specific feed/checkout work only when the site actually sells. For a content or docs site, treat articles, docs pages, and FAQ entities as the catalog and keep the door pointing at the real product (e.g. the app signup or the docs), not a fake shop.
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

