Questions are part of the evidence.

No. Catalume combines ground-truth retrieval where a platform exposes the same commerce substrate with high-fidelity provider simulations and clearly labeled modeled stages. It shows evidence and fidelity badges instead of pretending every replay is the consumer app.

Only public storefront pages and publicly reachable commerce surfaces. It does not ask for credentials, bypass a storefront password, or read shopper data. Detailed diagnosis and private store data require install or verified control.

Discoverability measures whether an agent can find the store and product. Comprehension checks whether it describes the facts correctly. Choosability checks comparison and recommendation. Transactability walks the path to checkout without completing payment.

No. Fixes are presented as reviewable artifacts. Nothing publishes without explicit approval, applied changes are revertible, and verification re-runs the failed mission plus collateral checks.

A grade is a time-bounded, opt-in summary of observed agent readiness. It is not a product-quality score, revenue claim, certification, or guarantee of visibility.

The product models ChatGPT, Gemini, Claude, Perplexity, and Copilot contexts. Each stage is labeled as retrieval, reasoning/choice, or checkout walk so the limits stay visible.

Plans scale by coverage volume, mission credits, automation depth, and store count. Verification is included on paid tiers because proof is part of the product, not a separate meter.

Email support@catalume.xyz for product questions, scanner issues, or a public-grade correction request.