This is a sanitized public demo of a private AI-powered customs classification assistant.
The best way to help is to choose one review lane and open a focused GitHub Issue.
Focus:
- React/TypeScript structure
- component boundaries
- data flow
- maintainability
- what would break when the app grows
- what should be simplified or deleted
Best feedback:
- "This pattern will not scale because..."
- "This state should move here because..."
- "This component has too many responsibilities..."
Focus:
- accidental data exposure
- unsafe assumptions
- public demo risks
- future Auth0/Convex/OpenAI risks
- frontend information leakage
Best feedback:
- "This would be dangerous in production because..."
- "Before public release, you need to test..."
- "Never expose this client-side..."
Focus:
- what should be covered by unit tests
- what should be covered by E2E tests
- what smoke tests matter
- failure modes
- regression risks
Best feedback:
- "Add a test that proves..."
- "This edge case will break..."
- "This should be a CI gate..."
Focus:
- clarity
- trust
- usability
- accessibility
- mobile layout
- whether the disclaimers are understandable
- whether the result page creates false legal confidence
Best feedback:
- "This wording is misleading..."
- "This layout hides the important warning..."
- "A customs user would expect..."
Focus:
- whether the demo wording overclaims
- whether HS/CN/TARIC/BiH distinction is clear
- whether "illustrative HS family" is safer than "HS code"
- whether the validation flow feels credible for customs professionals
Best feedback:
- "This would be challenged by customs because..."
- "This distinction is missing..."
- "This source wording is too strong..."
Focus:
- broken UI
- broken links
- responsive layout issues
- browser/device issues
- deployment issues
Please prioritize feedback that answers one of these questions:
- What would make this system unsafe to trust?
- What would block this from becoming production-grade?
- What architectural decision should be changed early?
- What security issue would be expensive to fix later?
- What wording creates legal or professional overclaim risk?
- What test would give the most confidence?
P0 — dangerous: security leak, legal overclaim, misleading classification confidence P1 — blocker: broken demo, broken core flow, major architecture risk P2 — important: maintainability, UX confusion, missing tests P3 — polish: wording, layout, minor improvement
Open a GitHub Issue using the matching template:
- Bug Report
- Architecture Review
- Security Review
- Testing Review
- UI/UX Feedback
- Customs-Domain Review
Please be direct. Critical feedback is useful.