← ClaudeAtlas

aeo-visibility-audit-prolisted

AEO/GEO (Answer Engine / Generative Engine Optimization) visibility audit. Invoke when you submit a project's page or site for analysis of whether AI answer engines (ChatGPT, Perplexity, Claude, Gemini) will cite it when asked about its category — and the specific gaps stopping it. Returns a citeability score, per-rule analysis, and a ranked fix list. This is the 2026 successor to SEO, built for crypto and tech. Authored by an ex-CMO of two international crypto exchanges running an AI-leveraged practice.
jukkablomberg/northpoint · ★ 1 · AI & Automation · score 77
Install: claude install-skill jukkablomberg/northpoint
# AEO / GEO Visibility Audit Pro — 12 Rules This is the engine behind `northpoint.fi/check/aeo`. The public version checks 4 rules; this skill checks all 12, with operator-grade fixes. The question this answers: when a founder, investor, or journalist asks an AI — *"what's the best L1 for RWAs?"*, *"which stablecoin rails are MiCA-ready?"* — **does the project get named, and cited, or does the model surface competitors instead?** Attention is moving from the ten blue links to the single synthesised answer. This audit measures readiness for that surface. --- ## Two modes 1. **Heuristic mode (default, works on submitted page text/HTML).** Analyses whether the *content itself* is structured for an answer engine to extract, attribute, and cite the project. This is what runs from page text or a fetched URL. 2. **Live mode (when engine access is configured).** Additionally queries the answer engines with the category prompts and reports actual citation position per engine. If live-engine results are not provided in the input, run heuristic mode and clearly state that the per-engine citation scores are *projected from content signals*, not live-measured. State which mode produced the result in the output. --- ## How to invoke this skill You submit: a URL or page text, the project's **category** in plain language, and (optionally) 2–5 **category prompts** a real user might ask an AI, plus competitor names. Optional live-engine transcripts. The model applies all 12 rules, sc