Open-source 26-criterion audit framework for measuring how well a website or API is understood by LLM agents (ChatGPT, Claude, Perplexity, Google AI Overview).
In 2025–2026, an increasing share of B2B SaaS and developer-tool discovery happens inside AI agents (Claude, ChatGPT, Perplexity, Google AI Overview), not Google. Sites that are not optimised for LLM consumption lose visibility quarter-over-quarter. Most "AI SEO" advice on the internet is generic; this repo gives you a concrete, scored checklist.
| File | Purpose |
|---|---|
audit-checklist.md |
26-criterion checklist with scoring (0–100), grouped into 5 categories. |
entity-authority-checklist.md |
Companion 13-criterion checklist for entity authority (50 points). For founder-led brands. |
prompts/audit-prompt.md |
Ready-to-paste Claude / ChatGPT prompt that runs the audit and produces a report. |
scripts/check-ai-readiness.sh |
Bash script that automates technical checks (~80% of category A + E). |
examples/ |
Sample audit reports for reference. |
- Open
audit-checklist.mdand run through 26 criteria. - Or paste
prompts/audit-prompt.mdinto Claude / ChatGPT with the target URL. - Get a 0–100 score with category breakdown.
./scripts/check-ai-readiness.sh https://example.comOutput: per-category score for criteria the script can detect (/llms.txt, /robots.txt policy, /sitemap.xml, JSON-LD presence, no-JS access, canonical, OpenGraph).
Total = sum across 5 categories (max 100)
≥75/100 → AI-ready (strong citation prospect)
50–74 → Workable, conversion gaps
30–49 → Significant gaps, AI agents miss content
<30 → Not citable, urgent fix needed
E1 gate: if content requires JavaScript to be visible, the whole audit is marked UNRELIABLE regardless of total — fix this first.
- You run a B2B SaaS and want to be cited in AI-generated answers for your category.
- You are an agency / consultant and want a structured deliverable for AI-optimisation engagements.
- You are evaluating a vendor and want to see how AI-ready their docs/API are.
- You are a founder preparing to launch and want to be discoverable from day one.
This checklist adapts and extends the methodology described in Gumeniuk.com — AI-ready API documentation (the 26-criterion framework) and incorporates entity authority signals based on observed LLM retrieval behaviour in 2025–2026.
Differences from the original 26-criterion framework:
- Re-weighted under B2B SaaS ICP (Category C — API spec — carries 25 pts).
- E1 gate ("content without JS") is a hard pre-condition, not a scoring criterion.
- Companion entity authority checklist (13 criteria, 50 pts) for founder-led brands where personal/organisation entity recognition matters.
- v0.1 — checklist + manual prompt
- v0.2 —
scripts/check-ai-readiness.shcovers all auto-detectable criteria - v0.3 — npm package
@ivannikov-pro/ai-readinessfor programmatic use - v0.4 — GitHub Action for AI-readiness CI gate
- v0.5 — Claude skill format for
~/.claude/skills/
Issues and PRs welcome. Particularly valuable: real-world audit reports for popular B2B SaaS (Stripe, Twilio, Linear, Notion, Vercel, etc.) — drop them in examples/ as anonymised case studies.
This is the open-source methodology. If you want someone to actually run the audit, implement fixes, and track AI Visibility Score over time — that's a paid service. See ivannikov.pro/en/services/ai-optimisation.
MIT — see LICENSE.
Aleksandr Ivannikov / IVANNIKOV.PRO — premium engineering team.
13+ years of shipping production systems for B2B SaaS, AI-native products, and Web3 teams.
- GitHub: @ivannikov-pro
- LinkedIn: an-ivannikov
- Telegram: @ivannikov_pro
- X / Twitter: @ivannikov_pro