What does this skill do, and when should you use it?
This skill is the repository’s dedicated Generative Engine Optimization (GEO) capability. It improves existing content for citation surfaces across ChatGPT, Perplexity, Google AI Overviews, Gemini, Claude, and Copilot by strengthening structure, factual density, authority signals, quotable language, and source attribution. It produces a ready-to-use asset or implementation-ready transformation with a short handoff summary. It is intended for GEO optimization, not structural on-page SEO audits or net-new drafting.
Reads the brief, target keywords, entity inputs, and quality constraints supplied by the user; analyzes definitions, quotable statements, factual density, citations, Q&A structure, authority signals, freshness, and clarity. When available, it consults canonical entity profiles and connected AI-monitoring or SEO tools. Its five-step workflow applies standalone definitions, sourced statements, expert or source signals, Q&A/table/list structures, specific data, and visible-content-matching FAQ schema. The output reports changes made, before-and-after GEO score, AI Query Coverage, metric provenance, and CORE-EEAT self-check results.
- A content team wants an existing article to become more citation-ready for ChatGPT or Perplexity.
- An SEO lead needs to optimize a page for AI Overviews, Gemini, or Claude without replacing a conventional on-page SEO audit.
- A brand team needs standalone, sourced answer blocks for a defined set of target AI queries.
- A site is losing clicks to AI Overviews and needs a measure-diagnose-rewrite-monitor recovery workflow.
How do you install this skill?
- Do not present the GEO score, Tavily citability, or structural improvements as guaranteed unprompted citations in ChatGPT, Perplexity, or Google AI Overviews.
- Before processing live URLs, brand entities, or third-party content, confirm authorization, robots.txt, privacy, and data-egress boundaries; treat missing entity profiles as DONE_WITH_CONCERNS.
- Recheck all statistics, platform behavior, crawler rules, and engine preferences against current sources before publication; the in-skill tables are heuristic.
- Confirm authorization, path safety, and recoverability before saving to memory or promoting findings to hot-cache or open-loop storage.
- Shell / CLI
- Network access
- Local filesystem
Python 3
For the full repository, run /plugin marketplace add aaron-he-zhu/aaron-marketing-skills in Claude Code, followed by /plugin install aaron-marketing@aaron; or run npx skills add aaron-he-zhu/aaron-marketing-skills on an Agent Skills-compatible host. You may also run git clone https://github.com/aaron-he-zhu/aaron-marketing-skills and use seo-geo/implement/geo-content-optimizer/SKILL.md.
tmp="$(mktemp -d)"
git clone --depth 1 https://github.com/aaron-he-zhu/aaron-marketing-skills.git "$tmp"
mkdir -p ~/.claude/skills
cp -R "$tmp/seo-geo/implement/geo-content-optimizer" ~/.claude/skills/
rm -rf "$tmp"Generated from the source repository and skill path; it copies only this skill's folder. If the author's install steps above differ, follow those first. To scope it to one project, replace ~/.claude/skills with that project's .claude/skills.
How do you use this skill?
Provide content or a URL and identify the target AI engine, for example: Optimize this content for GEO/AI citations: [content or URL]. Other supported prompts include making an article more likely to be cited, auditing GEO readiness, and creating an AI Overview recovery plan. Without connectors, provide target queries, content, engines, competitor examples, and known citation gaps.
What are this skill's strengths and limitations?
- Explicitly covers citation surfaces for ChatGPT, Perplexity, Google AI Overviews, Gemini, Claude, and Copilot.
- Uses a defined five-step workflow, output contract, provenance labels, and CORE-EEAT checks.
- Separates minute-scale citability tests from slower, uncertain unprompted surfacing caused by crawling and index refresh.
- Can work from user-supplied data; the repository is licensed under Apache-2.0.
- The supplied skill text points to, but does not include, the full GEO target tables, self-check details, and measurement protocol.
- The Tavily probe directly measures Tavily’s layer; its relevance to ChatGPT, Perplexity, and Google AI Overviews is only an estimated proxy.
- Live queries, URL fetching, and the Tavily probe require network access; saving results requires filesystem access.
- The example probe uses `${CLAUDE_PLUGIN_ROOT}` and `python3`, so other hosts may require path or execution adjustments.
How does this skill compare with similar options?
Side by side with related skills; every score comes from the same FSRS standard.
| Skill | FS score | Stars | Last updated | License |
|---|---|---|---|---|
| GEO Citation Optimizer this page | 52 · Use with care | ★ 3k | 12d ago | Apache-2.0 |
| SERP Analysis | 56 · Use with care | ★ 3k | 12d ago | Apache-2.0 |
| Blog GEO Citation Audit | 55 · Use with care | ★ 2.4k | 13d ago | MIT |
| GEO Citability Scoring Skill | 48 · Use with care | ★ 11k | 21d ago | MIT |
| AI Search & GEO Optimizer | 52 · Use with care | ★ 19k | 22d ago | MIT |
Compared with on-page-seo-checker, this skill targets AI citation readiness rather than structural on-page SEO. Compared with content-writer, it optimizes existing content or produces an implementation-ready transformation rather than serving as the repository’s net-new drafting skill.
How did FollowSkills review this skill?
The skill treats fetched content as untrusted data, distinguishes User-provided, Measured, and Estimated metrics, and requires confirmation before saving results; its connector guidance describes keyless, read-only probing. Points are deducted because memory writes, connector data flows, permission boundaries, rollback, and sensitive-content handling are not fully specified within the skill itself.
The five-step workflow, input/output contract, N/A rule, and DONE_WITH_CONCERNS state are reasonably coherent. Points are deducted because this is a static review with no key-path execution evidence or committed coverage tests, tool and reference availability is partly assumed, and failure diagnostics remain thin.
Triggers, intended tasks, argument hint, and boundaries with on-page-seo-checker and content-writer are explicit, with English and Chinese trigger wording. Points are deducted because engine-specific limits, mainland-China reachability, and URL-fetch failure cases are not fully defined, while some tactics depend on external tools or search indexes.
The skill provides front matter, version, Apache-2.0 licensing, quick starts, a contract, layered references, limitation disclosures, and a next-skill handoff. Points are deducted because maintainer ownership, update path, and changelog are not clear in the skill, troubleshooting is delegated to repository-level documents, and parameter/output edge examples are limited.
It specifies directly usable content, metadata, or schema deliverables, before/after GEO scoring, AI Query Coverage, and a self-check; it also correctly separates citability proxies from unprompted surfacing. Points are deducted because the core methods are heuristic templates, no representative outputs or third-party execution evidence are supplied, and results still require factual validation and human review.
The skill requires source-status labels and separates measured from estimated signals, while documenting different timescales for citability and surfacing and the need for control groups. Points are deducted because the supplied files contain no independently reproducible execution record, CI evidence covering this skill, or complete corroborating source material; key platform claims are explicitly pending validation.
Open a dimension to read why it scored that way
Evidence confidence：Low — Mostly static review, author material or a limited demo; useful for discovery, not high-risk decisions.
See the full review method →