← Ranking
How we score
Each store gets a score out of 100 from public signals only: what it declares to agents, what it documents for them, and whether they can reach it. We don't place orders or test payments. Last scan: 26 Oct 2026.
Our audit only reads public pages: the home page, robots.txt, agent files (llms.txt, agents.md, /.well-known/ucp, A2A agent card, MCP discovery), a few product pages for their structured data, and the home page requested as GPTBot, ChatGPT-User, ClaudeBot, Claude-User and Perplexity-User. Files generated by a platform (Shopify, BigCommerce…) count: what matters is what an agent can use. Stores whose site did not answer at all are left out.
| Protocols · 40 pts | |
| UCP profile | 15 |
| Agent checkout | 7 |
| Payment handlers | 5 |
| Agent API (MCP) | 5 |
| A2A agent card | 4 |
| ACP | 2 |
| Signed agent identity | 2 |
| Documentation · 20 pts | |
| llms.txt | 7 |
| llms-full.txt | 2 |
| agents.md | 4 |
| WebMCP | 3 |
| robots.txt | 2 |
| robots.txt speaks to AI agents | 2 |
| Product data · 20 pts | |
| Product structured data | 6 |
| Price | 4 |
| Availability | 4 |
| Product identifier | 3 |
| Shipping & returns | 3 |
| Access · 20 pts | |
| Homepage reachable | 6 |
| GPTBot | 2 |
| ChatGPT-User | 2 |
| ClaudeBot | 2 |
| Claude-User | 2 |
| Perplexity-User | 2 |
| AI crawlers allowed in robots.txt | 4 |
| Grades: A ≥ 80 · B ≥ 60 · C ≥ 40 · D ≥ 20 · E < 20 | |