FAQ

The big questions

How is E-GEO different from geo-optimizer-skill?

geo-optimizer-skill (Auriti-Labs) is a strong, popular tool — ~644 stars as of 2026-08 vs E-GEO’s 147 — offering a CLI, Python library, MCP, and an Astro integration, and scoring sites 0–100 across 47 methods. If you want the broadest audit coverage, it is a fair choice.

E-GEO differentiates on four things:

1. Rewriting, not just auditing — the full analyze → rank → rewrite → schema pipeline outputs copy-paste-ready optimized content.
2. Reproducible evaluation harness — measure prompt quality yourself, offline, deterministically; the same check runs in CI.
3. Continuous geo-loop mode — watch domains over time with a persistent workspace ($EGEO_HOME) and deterministic collectors.
4. Research-backed — built on the E-GEO paper (arXiv:2511.20867) and the Princeton GEO study (KDD 2024).

Full comparison: E-GEO vs geo-optimizer-skill.

Does E-GEO work without Claude Code?

Yes. The standalone egeo Python CLI runs the full pipeline anywhere Python runs — local shells, notebooks, Docker, CI:

pip install -e .
egeo optimize path/to/page.md

Claude Code adds the agent workflow (/geo slash commands), auto-triggered skills, and MCP-based validation, but it is not required. Both runtimes share the same geo_eval.py / llm_client.py modules — no duplicated logic. The one exception: the interpretive part of a loop-mode run (/geo:loop <domain>) is executed by an agent runtime; the LLM-free parts (egeo loop run|collect|doctor) work from any shell.

Do I need API keys?

Not to try it. Every command honors GEO_EVAL_MOCK=1, which swaps in a deterministic mock LLM client — no API key at all. For real runs:

Key	Needed for
OPENAI_API_KEY	Real model runs (any OpenAI-compatible endpoint via OPENAI_BASE_URL)
BRAVE_API_KEY	The serp collector in loop mode only

Is GEO the same as SEO?

No. They share the goal of visibility but target different systems:

Aspect	SEO	GEO
Target	Google, Bing crawlers	ChatGPT, Perplexity, Claude, Gemini
Key signals	Backlinks, keywords, page speed	Content features, authority, relevance
Outcome	Ranked blue links	Citations in AI-generated answers

GEO complements SEO — you should do both. Deep dive: GEO vs SEO.

General

What is Generative Engine Optimization (GEO)?

GEO is the practice of optimizing content to rank higher in AI-powered search engines like ChatGPT, Perplexity, Claude, and Gemini. Unlike traditional SEO (which optimizes for crawlers), GEO focuses on how Large Language Models perceive, understand, and recommend content. See What is GEO?.

Is E-GEO based on real research?

Yes. E-GEO is based on the paper “E-GEO: Optimizing Content for Generative Engines” (arXiv:2511.20867), which identified 10 universal features that consistently improve AI-engine rankings.

Do I need coding skills to use E-GEO?

No for the Claude Code workflow — you type commands like /geo https://yoursite.com. Basic terminal familiarity helps for the standalone CLI.

Usage

What can I optimize?

Local Markdown/text files via egeo optimize <file> or /geo:optimize <file>; any public webpage via /geo <url> in Claude Code (fetched through MCP fetch or a web-reader fallback); entire folders via /geo:batch.

What if I don’t have MCP servers configured?

E-GEO works without them, but outputs are marked “Low Confidence”. For best results configure Brave Search and Chrome DevTools MCP servers — the validation-doctor skill prints exact setup commands. See MCP Server.

Can E-GEO optimize content in languages other than English?

Currently E-GEO is optimized for English content. Multi-language support is on the roadmap.

Output & results

Are the ranking predictions guaranteed?

No. Ranking predictions are estimates based on simulated AI-engine behavior. Actual rankings depend on content quality, competition, AI engine algorithm changes, and time since optimization. The evaluation harness documents this limitation explicitly: its metrics are a proxy, not real ChatGPT/Perplexity rankings.

Does E-GEO fabricate statistics or testimonials?

No. E-GEO never invents statistics, testimonials, or ratings. If your content lacks social proof, the report flags it as a gap with recommendations on how to add real proof.

What schema types does E-GEO generate?

SoftwareApplication, Organization, Article, Product, Service, and FAQPage. The CLI’s --schema-type flag currently accepts Organization, Product, Service, Article, and FAQPage.

Where do I put the generated schema?

Add the JSON-LD to your HTML <head>:

<script type="application/ld+json">
{ ... generated schema ... }
</script>

Technical

What LLM does E-GEO use?

The Claude Code workflow runs on Claude. The evaluation harness and CLI can use any OpenAI-compatible API (OPENAI_BASE_URL), defaulting to gpt-4o for ranker, rewriter, and meta-optimizer models.

Can I customize the agents?

Yes. Agent configurations live in .claude/agents/ as markdown files with system instructions. Edit them to customize behavior.

Troubleshooting

“MCP servers unavailable” warning

Add Brave Search and Chrome DevTools MCP servers to your Claude Code MCP config, or run the validation-doctor skill for exact setup commands. E-GEO still works without them — outputs are just marked Low Confidence.

Optimization takes too long

Single pages typically take 2–5 minutes; batch folders 5–15 minutes. Try /geo:audit first (analysis only, faster), and check that the target URL is accessible.