zensink/zens-ink-seo-package/geo-readiness

GEO: get cited by AI engines

AI engines cite pages they can retrieve, parse, and trust. This skill covers the three layers: crawler policy (robots.txt), machine-readable index (llms.txt), and content shaped for fan-out queries.

When to use

• "How do I get my site cited by ChatGPT / Perplexity?"
• "Generate llms.txt for my site"
• "Audit my AI readiness"
• Planning content that answers AI-spawned sub-questions.

When NOT to use

• The site has zero conventional rankings — AI citation mostly mirrors SERP

strength; fix rankings first (keyword-research + site-audit skills).

Setup

pip install zens-ink

Commands

# 1. Audit the AI layer: robots.txt crawler policy + llms.txt presence
python3 -m zens_ink.ai_crawler_audit https://example.com
python3 -m zens_ink.ai_crawler_audit https://example.com --markdown report.md
 
# 2. Generate llms.txt + llms-full.txt from a build or remote sitemap
python3 -m zens_ink.llms_gen --dist dist --site https://example.com \
  --name "Example" --tagline "What the site does"
python3 -m zens_ink.llms_gen --sitemap https://example.com/sitemap.xml \
  --site https://example.com
 
# 3. Plan Query Fan-out content (sub-queries AI engines derive)
python3 -m zens_ink.geo_fanout "best crm" --icp "small business owners"
python3 -m zens_ink.geo_fanout "ai seo" --lang zh
 
# 4. Gate drafts for citability before publishing
python3 -m zens_ink.content_qc drafts/

How to read the results

• ai_crawler_audit: decide the stance deliberately — allow search/RAG

crawlers (ai-input=yes, get cited) while denying training crawlers (ai-train=no, protect original IP). Content Signals comments in robots.txt express this machine-readably.

• llms_gen: llms.txt is the index, llms-full.txt the deep layer. Keep

descriptions factual (what the page answers), not marketing copy — agents route by task, not by adjectives.

• geo_fanout: each sub-query is a heading your content should answer

directly (BLUF style, quotable facts). Fan-out coverage = more retrieval surface across engines.

• content_qc: scores whether a draft is citable (BLUF, structure,

quotable specifics). Below 70, fix the named dimension.

The hard truth: plumbing does not create citations — retrieval does. SERP presence feeds AI visibility. Treat this layer as necessary, not sufficient.

Next steps

• Track whether citations actually appear: run a weekly probe of buyer

questions against AI search stacks and diff the answers over time (pattern: fixed question set, saved answers, delta report).

• Technical issues blocking indexing → site-audit skill.

Going deeper

ZensInk Pro ships a 16-dimension GEO Score (fact layer, structure, semantic, AI-access, trust) with prioritized P0/P1/P2 fixes: https://zens.ink/pricing/