Rust Review

Let the toolchain do the mechanical pass first, then read for the things it can't see. Don't hand-review what clippy already flags.

1. Run the toolchain

cargo fmt --check                 # formatting
cargo clippy --all-targets --all-features -- -D warnings   # lints as errors
cargo test                        # tests pass
cargo build --release 2>&1 | tail # warnings

If cargo-audit / cargo-deny are installed, run them for advisories: cargo audit. Fix or justify every clippy finding — don't #[allow] silently.

2. Read for what clippy misses

Unsafe. Every unsafe block needs a comment stating the invariant that makes it sound, and a reason a safe alternative won't do. Unjustified unsafe is the highest-priority finding.

Panics in library code. unwrap() / expect() / panic! / indexing v[i] / integer division — fine in tests and main, suspect in library paths. Prefer ?, ok_or, get, checked_*. Flag each one: can this input make it panic?

Error handling. Libraries: typed errors (thiserror). Applications: anyhow/eyre with context (.with_context(...)). Errors should say what failed and why. No Result<_, String> in real code.

Ownership & borrowing. Needless .clone() to dodge the borrow checker; String where &str would do; Vec<T> args where &[T] works; returning owned data that could be borrowed. APIs should take the least-owning type that works.

Idioms. Iterator chains over manual index loops; if let / let else / matches! over verbose matches; impl Trait where it reads cleaner; derive (Debug, Clone, PartialEq) rather than hand-writing; From/TryFrom for conversions. Newtypes over bare primitives for domain types.

Async (tokio). No blocking calls (std::fs, std::thread::sleep, heavy CPU) inside async tasks — use the async equivalents or spawn_blocking. Watch for holding a std/parking_lot lock across .await (deadlock risk); don't .await while holding a Mutex guard. Check that spawned tasks are awaited or intentionally detached, and that cancellation is handled.

Concurrency. Arc<Mutex<…>> that could be a channel or RwLock; data shared without Send/Sync thought-through.

3. Tests

• Meaningful coverage of behavior and failure paths, not just the happy path.
• #[should_panic] / Result-returning tests where apt.
• Property tests (proptest) for parsers/encoders.
• Doc examples (cargo test --doc) compile and run.

Severity

• Blocker: unjustified unsafe, a reachable panic on user input, a .await-across-lock deadlock, UB.
• Important: missing error context, needless clones in hot paths, blocking in async, missing tests on a failure path.
• Minor: non-idiomatic but correct; naming; style clippy didn't catch.

Report findings as file:line + the concrete fix. Only flag what's worth the author's time.