Rust Agents Code Reviewer

This skill reviews already-written Rust code for quality, idiom, and latent bugs. It is the deep-review counterpart to [[rust-agents-orchestrator]] (which routes and runs a fast checklist) and [[rust-agents-compile-fix]] (which iterates on compiler errors).

ALWAYS use this skill when a prompt asks "review this Rust code", "is this idiomatic", "find code smells", or before approving a Rust pull request. NEVER treat compilation success as a passed review: cargo build proves type-checking, not quality.

This skill reviews. It does NOT route tasks (that is rust-agents-orchestrator) and does NOT fix compiler errors (that is rust-agents-compile-fix). Treating these three as interchangeable is an anti-pattern.

The Review Loop

ALWAYS follow this order. Each step is a hard gate.

1. Mechanical pass: run cargo clippy --all-targets --all-features -- -D warnings and cargo fmt --all --check. Resolve every clippy correctness and suspicious finding before human review continues.
2. Naming pass: check every identifier against the Rust API Guidelines casing table below.
3. Error-handling pass: locate every .unwrap() / .expect() / panic! and demand a justification for each.
4. Async pass: check every async fn and spawned future for blocking calls, locks held across .await, and Send bounds.
5. Memory pass: check every Arc<Mutex<T>>, RwLock, RefCell, and .clone() for appropriateness.
6. Suppression pass: check every #[allow(...)] for a reason comment; recommend #[expect(...)].
7. Verdict: produce the structured verdict from the review checklist in references/methods.md.