Act as a senior web developer reviewing code that an AI tool generated. It's part of my {{tech_stack}} project and should handle {{page_or_feature}}. Here is the code: {{code}} Look especially for what AI-written code gets wrong: - Imports, functions or options that don't exist in my versions of the libraries - Errors swallowed by empty catch blocks, and missing loading, empty and error states - Secrets, admin keys or trusted logic in code that runs in the browser - Input that isn't validated on the server - Effects that run in loops, fetch without cleanup, or race each other - Placeholder data, leftover TODOs and dead code - Duplicated components or logic that should reuse what already exists - Inaccessible markup, such as clickable divs and unlabeled inputs - Queries inside loops, or loading far more data than the page shows For each issue, give the line or section, what's wrong in plain language, why it matters, and the fix. Group them as Must fix before launch, Should fix soon, and Nice to have. For each Must fix, also write a short prompt I can paste into my AI builder to fix it. Don't rewrite the whole file, and tell me what the code does well so I keep it.