Scoring rubric
Pass earns the full weight, warn earns half, and fail earns zero. The category weights total 100.
AI Discovery
20 pointsPublic files and policies that let answer engines discover the site.
llms.txt is available
A concise, root-level content map gives answer engines an explicit discovery surface.
llms-full.txt is available
A bounded full-text companion gives agents richer context without scraping the entire site.
AI crawler policy is explicit
Explicit crawler rules are more trustworthy than assuming access from an absent policy.
MCP discovery manifest is valid
A valid MCP manifest advertises real, agent-operable capabilities.
Agent policy is published
A machine-readable agent policy defines safe public and protected behavior.
Crawlability
20 pointsSignals that let crawlers safely find, select, and index canonical pages.
robots.txt is available
robots.txt is the baseline crawl-policy surface.
XML sitemap is available
A valid sitemap provides bounded, explicit URL discovery.
robots.txt references the sitemap
The reference joins crawl policy and URL discovery.
Canonical URL is declared
Canonical selection reduces duplicate and conflicting source identities.
Page is indexable
A page cannot be cited consistently if page-level policy blocks indexing.
Structured Meaning
20 pointsMachine-readable semantics used for entities, previews, and API discovery.
JSON-LD is syntactically valid
Valid JSON-LD is the strongest page-level machine-readable entity signal.
JSON-LD identifies useful entity types
Typed entities are more useful than syntactically valid but untyped graphs.
Open Graph metadata is complete
Stable title, description, image, and type metadata improve source previews.
Social card metadata is available
A card declaration supports consistent previews across secondary discovery channels.
Machine-readable API schema is available
OpenAPI or MCP discovery lets agents verify available operations instead of guessing.
Answer Quality & Trust
25 pointsContent, attribution, and freshness signals that support defensible citations.
Title is descriptive
A concise title establishes the page’s primary answer context.
Description summarizes the page
A useful description provides a compact statement of intent and scope.
A single primary heading is present
One clear H1 gives parsers a stable primary topic.
Content has useful depth
Substantive visible text is more likely to contain answerable, citable detail.
Content exposes answer structure
Question headings, lists, tables, and FAQ markup make answer units easier to extract.
Authorship or publisher is attributable
Attribution supports provenance and confidence in a citation.
Claims link to supporting sources
Outbound source links provide inspectable provenance for factual claims.
Freshness is machine-readable
Published or modified dates help engines reason about time-sensitive claims.
Reach & UX
15 pointsSyndication, locale, transport, mobile, and response-time readiness.
Document language is declared
Language metadata improves parsing, accessibility, and locale selection.
RSS or Atom feed is available
Feeds provide durable, incremental discovery for changing content.
Mobile viewport is configured
Mobile usability affects both readers and mobile-first indexing.
Primary HTML responds promptly
Fast server response improves crawl efficiency and user access.
Final page uses HTTPS
Secure transport is a baseline trust and integrity requirement.