Rust project setup

Rust Project Setup

Rule

Scaffold a brand-new Rust project; do not only propose a structure. Layer Rust-specific conventions on top of base-repo-setup and target greenfield work only.

Greenfield Rust should be Cargo-first, safe by default, explicit about MSRV and features, and dependency-minimal until a real domain need exists.

Hard Stops

Stop before writing files when:

• The repository baseline is missing (flake.nix, mise.toml, justfile, .gitignore).
• The target already has Cargo.toml, Cargo.lock, src/, tests/, or files that would be overwritten.
• Crate name, app/library shape, workspace shape, edition, MSRV, license, or publish intent is unclear.
• The request is a migration of an existing Rust project; these skills are greenfield-only.
• Setup would introduce async runtime, HTTP framework, database, config framework, OpenTelemetry, unsafe code, FFI, or release automation without approval.

Required Decisions

Ask and confirm:

1. Crate or workspace name. Use kebab-case package names and snake_case library crate names.
2. Shape: binary app, library crate, mixed app+lib, or workspace.
3. Edition. Recommend Rust 2024 when supported by the chosen stable toolchain.
4. MSRV. Libraries need explicit MSRV; apps may track stable unless release policy says otherwise.
5. License, description, repository URL, and publish intent.
6. Whether CLI, HTTP API, async, database, config, observability, benchmarks, docs examples, or release automation are in scope. Default is no; use specialized skills later.

Greenfield Defaults

• Use stable Rust. Pin toolchain through repo policy when the project requires repeatability.
• Prefer cargo new --bin or cargo new --lib shape, then adjust deliberately.
• Use src/main.rs for binaries, src/lib.rs for libraries, and tests/ for integration tests.
• Add a workspace only when multiple crates are requested.
• Track Cargo.lock for apps and binaries. Discuss lockfile policy for libraries and workspaces; default to tracking it in this personal baseline for reproducibility.
• Keep dependencies empty initially unless generated code imports them.
• Use rustfmt, Clippy, cargo test, cargo test --doc, and cargo doc --no-deps in validation.
• Use cargo-nextest only when approved or already standard; cargo test remains the portable baseline.
• Use cargo-deny for license/advisory policy and cargo-audit only when the repo chooses it; do not add both without need.

Just Recipes

Keep just check as the canonical validation entrypoint. Prefer recipes like:

fmt:
    cargo fmt --all

fmt-check:
    cargo fmt --all -- --check

lint:
    cargo clippy --all-targets --all-features -- -D warnings

test:
    cargo test --all-targets --all-features

doc:
    cargo doc --no-deps --all-features

check: fmt-check lint test doc

Add nextest, deny, audit, coverage, or bench recipes only when those tools are approved and versioned through project tooling.

Workflow

1. Verify the baseline repository files and Git root.
2. Ask all required decisions.
3. Create Cargo metadata, source skeleton, smoke tests, README updates, and just recipes.
4. Add .gitignore entries for generated Rust artifacts only when missing (target/, coverage/profiling outputs). Do not ignore Cargo.lock without approval.
5. Run cargo fmt, cargo test, cargo clippy, cargo doc, and just check.

Completion

Report crate/workspace shape, edition/MSRV, lockfile policy, files created, dependencies added or avoided, commands run, validation results, and intentionally skipped optional items such as Tokio, Axum, Clap, SQLx, OpenTelemetry, CI, and release automation.