Language focus
AI code review for Python
Python rewards fast iteration, which also means dynamic edges: mutable defaults, import-time side effects, dataclass mutability, and subtle async lifetime bugs that read fine at a glance.
Where CodeCritic concentrates for Python
- Exception handling that swallows context, fragile `try`/`finally` ordering, and logging that leaks PII.
- SciPy-heavy stacks mixing NumPy dtypes and implicit broadcasts - correctness issues reviewers miss until production.
- Packaging ambiguity (implicit namespace packages versus legacy layouts) hurting reproducible deploys.
- Security seams: deserialization, subprocess usage, tempfile races, permissive Flask/Django configs.
Regulated environments still need retention, access, and data-flow policies from your compliance team beyond what any reviewer - human or automated - summarizes on a ticket.