| name | rust-code-review |
| description | Use this skill when reviewing Rust code, diagnosing borrow checker errors, checking unsafe code, improving Cargo feature boundaries, or evaluating performance-sensitive Rust changes. Also use it when the user asks for a Rust review, Rust bug hunt, compiler error explanation, or refactoring guidance. |
Rust Code Review
Workflow
- Start with behavioral correctness: ownership, lifetimes, trait bounds, panic paths, integer bounds, UTF-8 boundaries, IO errors, and concurrency assumptions.
- Check API compatibility next: public type changes, feature gates, target-specific code, error messages, and default behavior.
- Review performance only after correctness: unnecessary allocation, repeated filesystem scans, lock scope, clone-heavy paths, avoidable parsing, and hot-loop formatting.
- If code was changed, ask for or infer the verification surface: unit tests, feature-specific checks, target-specific checks, and smoke tests for CLI/server paths.
Output
Lead with findings. For each finding, include:
- severity
- file/function or code area
- concrete failure mode
- minimal fix direction
If no major issue is visible, say so and mention remaining test gaps.
Gotchas
- Do not recommend broad rewrites when a local fix preserves existing behavior.
- Treat
unsafe, std::fs, network, subprocess, and lock-order changes as higher risk.
- For CLI tools, verify both one-shot and long-running modes when state is added.
- For feature-gated Rust, check at least the default feature set and the narrow feature set touched by the change.
