30 checks in Content Quality & Entity Semantics. Automated Its subsection has a live automated checker in the engine.
| ID | Check | Passes when |
|---|---|---|
| 3.10.01 | Check if site publishes an llms.txt file at the root domain | File present at /llms.txt |
| 3.10.02 | Validate llms.txt is well-structured and follows emerging standard format | Contains purpose, capabilities, content summary sections |
| 3.10.03 | Validate llms.txt accurately reflects actual site content and services | Descriptions match real site offerings |
| 3.10.04 | Validate llms.txt includes clear business entity identification | Business name, industry, primary offerings present |
| 3.10.05 | Validate llms.txt declares content limitations or scope boundaries | Limitations section present |
| 3.10.06 | Validate llms.txt is served with correct MIME type and encoding | text/plain; charset=utf-8 |
| 3.10.07 | Validate llms.txt file size is within AI context window limits (< 8k tokens) | Under 8k tokens |
| 3.10.08 | Validate llms.txt is accessible without authentication or bot-blocking | HTTP 200, no CAPTCHA |
| 3.10.09 | Validate llms.txt includes sitemap or content index references | References present |
| 3.10.10 | Compute LLMs.txt Presence & Quality composite score | Finalized |
| ID | Check | Passes when |
|---|---|---|
| 3.10.11 | Check if site exposes Model Context Protocol (MCP) endpoints or references | MCP endpoint or reference discoverable |
| 3.10.12 | Validate MCP endpoint responds with valid tool/resource definitions | Valid MCP schema |
| 3.10.13 | Validate MCP tool descriptions are clear, actionable, and accurate | Descriptions machine-parseable |
| 3.10.14 | Validate MCP endpoints are versioned and documented | Version info present |
| 3.10.15 | Compute MCP Discoverability composite score | Finalized |
| ID | Check | Passes when |
|---|---|---|
| 3.10.16 | Measure primary business information density within first 4k tokens of key pages | Core entity info within first 4k tokens |
| 3.10.17 | Validate key pages deliver complete context without requiring deep multi-page crawling | Single-page context sufficient |
| 3.10.18 | Validate page content is free of boilerplate bloat that wastes AI context window space | Content-to-boilerplate ratio > 60% |
| 3.10.19 | Validate critical information is not buried below the fold or behind expandable sections | Key info visible in initial render |
| 3.10.20 | Validate structured data is concise enough for AI context windows (no schema bloat) | Schema within reasonable token count |
| 3.10.21 | Validate meta description + title + H1 form a coherent context snapshot | Coherent triple |
| 3.10.22 | Validate AI-relevant content is not obscured by cookie banners, modals, or popups | Clean initial content extraction |
| 3.10.23 | Compute Context Window Optimization composite score | Finalized |
| ID | Check | Passes when |
|---|---|---|
| 3.10.24 | Validate content uses clear headings and concise paragraphs navigable by AI agents | H1-H6 hierarchy with short paragraphs |
| 3.10.25 | Validate content sections are machine-parseable with consistent formatting patterns | Consistent section structure |
| 3.10.26 | Validate presence of structured FAQ, how-to, or Q&A blocks extractable by AI | Extractable structured blocks present |
| 3.10.27 | Validate code examples, formulas, or technical content uses proper markup (pre, code, math) | Proper semantic markup |
| 3.10.28 | Validate table data uses proper HTML table elements (not div-based fake tables) | Semantic table markup |
| 3.10.29 | Validate page language attribute matches actual content language | lang attribute accurate |
| 3.10.30 | Compute AI Agent Content Structure composite score | Finalized |
← 3.9 Information Gain & Novelty Evaluation Suite3.11 Documentation Structure Quality →
A Deep Audit scores every check here that applies to your page, then an AI pass of up to 150 checks. Included on Pro and Ultra, or $9 for one audit.
See plansMachine-readable: catalog totals and the full catalog as JSON (IDs, section, weight, status).