Methodology
Every check is scored pass, warn or fail. Pass earns full weight, warn earns half, fail earns zero. The weighted total is normalized to a 0 to 100 score, then graded: A is 90 and up, B is 75, C is 60, D is 40, F below that. Here is every check and what it is worth.
Agent files
| llms.txt present and valid | 12 | The file agents read first: a markdown map of your key content. |
| Agent skill file (skill.md) | 8 | Tells an agent what your site does and how to use it. |
| MCP manifest | 6 | Lets agents call your service as a tool. |
| llms-full.txt present | 5 | The full-text version of your key content for LLMs. |
| ai.txt / ai-plugin | 2 | Optional, legacy AI usage manifest. |
Crawler access
| AI crawlers allowed | 14 | Whether robots.txt lets GPTBot, ClaudeBot, PerplexityBot and OAI-SearchBot in. The heaviest single check: if they are blocked, nothing else matters. |
Structured data
| Structured data (JSON-LD) | 10 | schema.org data that tells a machine what a page is about. |
| Meta description | 5 | A short summary of the page. |
| OpenGraph tags | 4 | So shares unfurl and some agents parse the page. |
| Page title | 4 | A clear, descriptive title. |
Content & discovery
| Content in initial HTML | 10 | Real text present before JavaScript runs, since many agents do not run it. |
| XML sitemap | 6 | A sitemap at the root or declared in robots.txt. |
| Machine-readable feed | 3 | Optional RSS or Atom feed. |