Review AI-generated code before I merge it

Review code that was written with an AI coding assistant before I merge it. The code or diff: [PASTE] What I asked the assistant to do: [THE ORIGINAL TASK OR PROMPT] Language, framework, and versions: [e.g. "TypeScript, Next.js 15, Prisma 6"] Project conventions: [ERROR HANDLING STYLE, LOGGING, FOLDER STRUCTURE, EXISTING HELPERS - optional] How much I understand this code: [FULLY / PARTLY / NOT REALLY] Check specifically for problems that are common in generated code: 1. Invented or outdated APIs: functions, options, or imports that may not exist in my versions. List each one as "[VERIFY in docs]" rather than assuming. 2. Scope drift: changes beyond what I asked for, deleted or weakened code, disabled checks, removed error handling, or commented-out tests. 3. Tests that don't test: assertions that always pass, mocks that replace the code under test, tests that mirror the implementation instead of checking behavior. 4. Security: unvalidated input, injection, missing authorization checks, secrets in code, overly permissive CORS or file access, unsafe deserialization. 5. Silent failure: swallowed exceptions, fallback values that hide errors, retries without limits. 6. Duplication: reimplementing helpers that likely exist in my codebase, or adding new dependencies unnecessarily. 7. Edge cases: empty inputs, nulls, large inputs, concurrency, time zones. Return: - A verdict: merge, merge after fixes, or rewrite. - A findings table (location, issue, severity, fix). - The questions I should be able to answer about this code before I own it in production. If I said I don't fully understand it, explain the parts I most need to understand.