A Claude Code skill for structured technical audits of Rust projects, adapted from the VGV Flutter Code Assessment methodology.
VGV's Flutter assessment has proven that repeatable, opinionated audits catch architectural drift before it becomes debt. Rust's compiler catches memory and concurrency bugs, but it doesn't enforce architectural discipline — a Rust project can compile cleanly while suffering from monolithic crates, concrete types everywhere, .unwrap() in production paths, and no CI gates.
This skill applies VGV's battle-tested principles (layered architecture, testability through injection, deterministic state, enforced quality gates) translated into Rust-native idioms.
Produces a RUST_ASSESSMENT.md with 10 sections covering:
- Architecture & Crate Structure — workspace layout, trait-based DI, layered separation
- Error Handling —
thiserror/anyhow, documented error conditions, no silent failures - Type Safety & Correctness — newtypes, pattern matching, unsafe audit
- Async & Concurrency — tokio patterns, cancellation safety, shared state
- Testing — coverage (≥90%), isolation, descriptive names, behavior-driven
- CI/CD & Tooling — clippy, fmt, deny, audit, conventional commits, semver
- API Design & Documentation — visibility, rustdoc, Diataxis,
# Errorssections - Dependencies — supply chain, bloat, compile time
- Performance & Resource Management — allocation, I/O, observability
- Refactoring Estimation — top risks, tasks, time estimates
/rust-code-assessment <target_directory>
Add to your Claude Code skills (personal or project level):
# Personal level (all projects)
cp -r .claude/skills/rust-code-assessment ~/.claude/skills/
# Project level (single project)
cp -r .claude/skills/rust-code-assessment your-project/.claude/skills/| VGV Flutter Standard | Rust Equivalent |
|---|---|
| Feature-first directories | Crate-per-concern workspace |
| BLoC (state separation) | State machines + channels |
| Repository pattern (DI) | Trait-based injection |
very_good_analysis |
clippy::pedantic + -D warnings |
| 100% test coverage | cargo-llvm-cov ≥90% |
mocktail |
mockall or manual test doubles |
| Conventional Commits | Conventional Commits |
| Semantic Versioning | Semantic Versioning |
MIT