Seller tool / Product operations
Verify the tools before building more tools.
Check each published SellerMind tool page, its controls, its linked API contract, and its AI/OpenAPI discovery entry. Use the result as a release gate for the self-optimization loop.
Run the smoke audit
First-party tool surface
The current deployment will be checked when you run the audit.
Decision rule
Repair drift before expansion.
A page can still rank and look polished while its API reference disappears, a control ID changes, or a contract is no longer published. Fix one failed check, run the same audit again, and keep the result as release evidence.
- Use the failed tool route as the first repair target.
- Keep one page, one contract, and one verification pass in the same change.
- Use browser automation only in an isolated worker when deep interaction evidence is justified.
Questions
Function audit FAQ
Does this click every tool?
No. This first contract checks page reachability, expected controls, references, and AI/OpenAPI coverage. Playwright-based rendered interaction checks are a separate future worker.
Why check the current origin only?
The product must first prove its own loop without turning a public endpoint into an arbitrary network scanner. Seller storefront scans remain under the separate public-target audit boundary.
Can it edit my storefront?
No. It only reads the deployed SellerMind surface. Storefront edits remain seller-authorized, human-reviewed, and outside this endpoint.