# Rust > Use when the project uses Rust for systems programming, CLI tools, WebAssembly, or backend services - Skill: `tomevault-io/rust-7` (Agent Skill, multi-file: 2 files) - Install (CLI): `npx skillmds@latest add tomevault-io/rust-7` - Raw SKILL.md: https://api.skillmd.com/api/skills/tomevault-io/rust-7/raw - Safety review: pending (external: skill-scanner PASS, skillspector PASS) - Works with: Claude Code, Claude.ai, OpenAI Codex - Category: Coding & Dev Tools - Author: tomevault-io (https://skillmd.com/u/tomevault-io) - Updated: 2026-09-17 - Page: https://skillmd.com/skills/tomevault-io/rust-7 --- # Rust Best Practices ## Ownership and Borrowing - Follow the ownership model: each value has a single owner, dropped when the owner goes out of scope - Use `&T` for borrowed references; prefer `&self` over `&mut self` in method signatures - Apply `Clone` only when necessary -- prefer references or `Copy` for small types - Use `Rc` for shared ownership in single-threaded contexts, `Arc` for multi-threaded - Avoid `RefCell` in performance-critical paths; restructure data to satisfy the borrow checker statically ## Error Handling - Use `Result` as the standard return type for fallible operations; never panic in library code - Define domain errors with `thiserror` for library crates: `#[derive(Error)]` with descriptive variants - Use `anyhow` for application-level error propagation with context: `.context("failed to open config")?` - Create a unified `AppError` type at application boundaries with `From` trait impls for conversion - Use `unwrap()` and `expect()` only in tests or cases where failure is genuinely impossible ## Async with Tokio - Use `tokio` as the async runtime; annotate `main` with `#[tokio::main]` and select the appropriate flavor (`flavor = "current_thread"` for lightweight, `multi_thread` for CPU-bound) - Prefer `tokio::spawn` for concurrent tasks; use `JoinHandle` to await results - Use `tokio::sync::Mutex` (not `std::sync::Mutex`) across `.await` points to avoid holding OS locks across yields - Apply `tokio::select!` for racing futures and handling cancellation - Use channels (`mpsc`, `broadcast`, `watch`) for inter-task communication over shared mutable state ## Trait Design - Design traits for behavior, not data; keep trait methods focused and composable - Use associated types when the return type varies by implementor: `type Output` - Apply trait bounds at the function level, not the struct level, to maximize flexibility - Implement `From`/`Into` for type conversions rather than ad-hoc methods - Use `dyn Trait` for dynamic dispatch only when static dispatch with generics is impractical ## Cargo and Project Structure - Use cargo workspaces for multi-crate projects: define `[workspace]` in the root `Cargo.toml` - Keep binary, library, and test targets separated under `src/bin/`, `src/lib.rs`, and `tests/` - Pin dependency versions with `Cargo.lock` for binaries; omit it for published libraries - Use `#[cfg(test)]` modules for unit tests within source files and `tests/` directory for integration tests - Profile for release builds: set `opt-level`, `lto = true`, and `codegen-units = 1` in `Cargo.toml` for maximum performance ## WebAssembly Targets - Target `wasm32-unknown-unknown` for browser WASM; use `wasm-bindgen` for JS interop - Use `wasm-pack` for building and publishing WASM packages - Avoid `std::fs`, `std::net`, and blocking I/O in WASM targets -- use browser APIs via `web-sys` - Keep WASM modules small: enable `wee_alloc` as the global allocator and tree-shake with LTO - Test WASM logic in native Rust first, then verify in browser with `wasm-pack test --chrome` --- > Source: [calcosmic/Aether](https://github.com/calcosmic/Aether) — distributed by [TomeVault](https://tomevault.io).