← ClaudeAtlas

aeo-auditlisted

Audit a website for AI Engine Optimization (AEO) — whether AI assistants and their crawlers can reach, parse, and cite it. Checks robots.txt AI-bot rules, llms.txt, JSON-LD structured data, sitemaps, canonical/meta signals, and .well-known endpoints, then reports findings by severity with concrete fixes. Use this whenever the user asks about AI search visibility, ChatGPT/Claude/Perplexity citations, AI crawler access, llms.txt, robots.txt bot rules, schema markup, structured data, "why doesn't AI mention my site", agentic commerce readiness, or asks to audit or review a site or URL for AI or LLM discoverability — even if they don't say "AEO".
angeo-dev/skills · ★ 1 · AI & Automation · score 72
Install: claude install-skill angeo-dev/skills
# AEO Audit Audit how a website looks to AI assistants and their crawlers, then report what to fix. Traditional SEO asks whether a page ranks. AEO asks something different: can an AI system fetch the page at all, parse what it says, and cite it with confidence? Those are separate failure modes. A site can rank on page one and still be invisible to an assistant because its robots.txt blocks the crawler, or because the only place a price exists is in JavaScript that never runs. ## Scope and honesty State what was measured, not what it implies. Two rules matter more than any individual check: **Never guess at a signal you could not fetch.** If robots.txt returns a 500, that is a finding — "could not verify", not "no restrictions found". Silent failures that read as passes are the fastest way to make an audit worthless. **Do not claim to know how any specific AI system ranks or selects sources.** Nobody outside those companies does. What is verifiable: whether a documented crawler is allowed, whether a machine-readable format is present and valid, whether facts are stated in the HTML rather than only in script. Report those. Frame everything else as a hypothesis. ## Workflow ### 1. Establish the target Take the URL the user gave. If they gave a bare domain, use `https://` and follow redirects — note if the canonical host differs from what they typed (`example.com` → `www.example.com`), since that mismatch is itself a common source of split signals. If they gave no URL,