← ClaudeAtlas

rust-code-reviewlisted

Review Rust changes on two axes at once — standards (idioms, ownership, errors, API surface, unsafe) and spec (does the change do what was asked) — with a Rust smell baseline layered on the repo lint configuration. Use when reviewing a Rust diff, pull request, or branch, when asked whether a change is ready to merge, or when a review needs to cover more than what clippy already reports.
rewrite-rs/skills · ★ 2 · Code & Development · score 73
Install: claude install-skill rewrite-rs/skills
# Rust Code Review A review is a statement about what a change owes — to the request that asked for it, and to the standards the repo already lives by. This skill runs both statements as separate passes and merges them into one report led by a verdict. It never restates a rule a smell maps to: the craft skills and `/rust-testing` own the standards, and the review routes to them. ## Two axes, run separately A single reviewer reading for both correctness-against-the-request and craft does neither well: spec drift hides behind clean-looking code, and style nits crowd out the missing edge case. Run a **standards pass** and a **spec pass** as separate passes — as parallel sub-agents when the harness supports it, sequentially when it does not — then merge. The prompts for both, the merge rules, and the report shape are in `REVIEW-PASSES.md`. ## The machine goes first Before either pass reads a line, run the verification step below and read the output. Anything clippy already reports is not a review finding; it is a build failure someone forgot to run. Reviewer attention is for what the tools cannot see: the judgment call, the missing edge case, the scope that crept in. ## Read the repo lint configuration before judging style `[lints.clippy]` in `Cargo.toml`, a `clippy.toml`, a `rustfmt.toml`, `#![deny(...)]` in the crate root. A finding that contradicts a level the repo deliberately set is not a finding — it is a proposal, and it says so explicitly. Never re-run at a stricte