Rubric r1, 25 checks
How the audit scores your store
Rubric r1 has 25 checks worth 100 points in six dimensions. Software measures 23 of them directly. An AI model judges the other 2, and every answer it gives has to quote the text on your pages it relied on.
Scores and grades
Each check earns a share of its points. Checks that don't apply to your store, or that couldn't run, leave the total, so your score is the share of the remaining points you earned, out of 100.
Grades: A is 85 to 100, B is 70 to 84, C is 55 to 69, D is 40 to 54, F is below 40.
The checks
Every store is scored on the same checks, each worth the points shown. Checks marked as judged by an AI model are answered from your page text, and an answer counts only when its quote matches the page.
Access, 20 points
| Check | What it looks at | How | Points |
|---|---|---|---|
| A1 | AI agents allowed by robots.txt | Measured by software | 6 |
| A2 | Honest crawler not blocked | Measured by software | 5 |
| A3 | Products not hidden behind a wall | Measured by software | 4 |
| A4 | Product data without JavaScript | Measured by software | 5 |
Discovery, 15 points
| Check | What it looks at | How | Points |
|---|---|---|---|
| D1 | Sitemap coverage | Measured by software | 4 |
| D2 | Agent profile published (UCP) | Measured by software | 5 |
| D3 | Catalog answers an agent's search | Measured by software | 3 |
| D4 | llms.txt present and valid | Measured by software | 1 |
| D5 | Canonical and crawlable product URLs | Measured by software | 2 |
Product data, 25 points
| Check | What it looks at | How | Points |
|---|---|---|---|
| P1 | Product structured data present | Measured by software | 6 |
| P2 | Recommended product fields | Measured by software | 5 |
| P3 | Variants modeled | Measured by software | 4 |
| P4 | Structured data matches the page | Measured by software | 4 |
| P5 | Product pages answer buyer questions | Judged by an AI model, which quotes its source | 6 |
Policies and trust, 15 points
| Check | What it looks at | How | Points |
|---|---|---|---|
| T1 | Return policy in structured data | Measured by software | 4 |
| T2 | Shipping details in structured data | Measured by software | 4 |
| T3 | Policies answer buyer questions | Judged by an AI model, which quotes its source | 4 |
| T4 | Business identity | Measured by software | 3 |
Checkout path, 15 points
| Check | What it looks at | How | Points |
|---|---|---|---|
| C1 | Agent checkout advertised | Measured by software | 4 |
| C2 | Add to cart works for agents | Measured by software | 4 |
| C3 | Cart link resolves (full report, with your authorization) | Measured by software | 4 |
| C4 | Guest checkout (full report, with your authorization) | Measured by software | 3 |
Agent experience, 10 points
| Check | What it looks at | How | Points |
|---|---|---|---|
| X2 | Page size within agent budgets | Measured by software | 4 |
| X3 | Accessible page structure | Measured by software | 5 |
| X4 | Markdown on request | Measured by software | 1 |
Each check links to its guide: what it measures, why agents care and how to fix it. The agent-readiness checklist has all of them on one page.
What we read
Our crawler, GlassAisleBot, fetches up to 40 public URLs from your store and loads up to 10 of them in a headless browser. It follows your robots.txt, never logs in, submits a form or places an order, and stops when bot protection turns it away. Every finding in a report cites the stored page, screenshot or code it came from.
See how agents read your store.
Free. No login or app install. Report in about 5 minutes.