SEO + GEO Playbook
An evidence-backed toolkit for SEO and GEO (Generative Engine Optimization), shipped as Claude Code skills. It audits a site the way an AI answer engine actually reads it, scores it A to F, drafts your llms.txt, and tracks whether ChatGPT, Perplexity and Gemini really cite you. Zero dependencies in the core, so the audit runs in seconds.
AI citations 13% to 30%+ in 90 days
What it is
An open-source playbook (github.com/OndrejKnedla/seo-geo-playbook-ok) for ranking in Google and getting cited by AI answer engines. It ships as a set of Claude Code skills plus a zero-dependency audit you can run with one command.
How it works
- seo-geo-audit crawls the server HTML (not the browser DOM, because the two differ and that gap is where visibility quietly dies), runs ~28 deterministic checks plus a 6-dimension citability rubric, and returns an A to F grade with a prioritized fix list.
- generate-llms-txt drafts a curated
/llms.txtfrom your sitemap. - track-ai-citations measures whether ChatGPT, Perplexity and Gemini actually cite you, re-tested over time.
- seo-geo-fix applies one focused fix (SSR JSON-LD, canonical, schema, headings) as a small reviewable PR.
- Plus four agents, a GitHub Action that gates deploys on the score, and an optional DataForSEO toolkit in Python.
The key decision
On-page SEO and GEO can be maxed out in weeks, then they plateau. The real ceiling is your off-site brand-mention footprint. Brand mentions correlate with AI visibility about 3x more strongly than backlinks (Ahrefs, Dec 2025). The playbook is built around that one idea: fix the page fast, then earn independent mentions.
Outcome
- AI-citation rate on a fixed 30-query set: ~13% baseline, ~22% after on-page work, 30%+ after the off-site footprint.
- Foundational SEO health: 64 to maxed after fixes.
- Open source, MIT, zero core dependencies.