rust-best-practices

name	rust-best-practices
version	0.11.0
description	Review Rust architecture plans, crate boundaries, and code for structural design-pattern compliance. Use when the task involves typestate, sealed traits, error contracts, wrapper/newtype design, object safety, interior mutability, or other type-system-driven Rust correctness patterns that go beyond general style guidance.
depends_on	{"rust-best-practices-agent":"0.x","rust-architect":"0.x","rust-code-reviewer":"0.x","rust-code-explorer":"0.x"}

Rust Best Practices

This skill is the canonical source of truth for structural Rust pattern review in raptor. It complements rust-development by focusing on type-system-driven correctness, API-boundary design, and the lifecycle stage where each pattern should be enforced.

Scope

Use this skill for:

• architecture and design review of Rust plans
• crate-boundary and public API review
• code review for structural Rust pattern compliance
• QA review when the goal is to enforce documented best-practice patterns

This skill does not cover service-runtime hardening such as timeouts, graceful shutdown, request IDs, retries, readiness probes, or backpressure. Those belong in rust-service-hardening.

Canonical Inventory

Read these files in order:

• patterns/practice-inventory.md — canonical practice ids, names, lifecycle stage, and review applicability
• patterns/enforcement-strategy.md — enforcement heuristics, stage mapping, and review guidance

Read these additional pattern docs only when the practice under review requires them:

• patterns/error-context-recovery-plan.md
• patterns/typestate-plan.md
• patterns/sealed-traits-plan.md
• patterns/newtype-zero-cost-plan.md
• patterns/deref-coercion-plan.md
• patterns/interior-mutability-plan.md
• patterns/infallible-plan.md
• patterns/trait-object-safety-plan.md
• patterns/cow-plan.md
• patterns/phantomdata-capability-token-plan.md

Review Modes

Design Review

Use for plans, architecture documents, and implementation blueprints.

Focus on:

• error contracts and recovery inventory
• typestate opportunities
• sealed traits and crate extension boundaries
• newtype / deref / zero-cost wrapper design
• trait object safety when plugin-style dispatch is intended
• capability-token / PhantomData patterns for resource invariants

Code Review

Use for implemented Rust code.

Focus on:

• missing structured error context and recovery guidance
• repeated primitive validation that should become newtypes
• unsafe or weakly justified interior mutability
• Result<T, E> shapes that should be simplified or made Infallible
• unnecessary ownership on hot paths where Cow fits
• wrappers whose ergonomics justify Deref/AsRef/Borrow

Crate Boundary Review

Use when extracting crates or defining public traits and extension points.

Focus on:

• sealed traits
• trait object safety
• wrapper/newtype boundaries for semantic ids and validated values
• which patterns must be enforced before downstream API exposure

Agent Delegation

This skill delegates pattern review work to existing Rust agents when specialized execution is needed:

Operation	Agent	Returns
Dedicated structural pattern review	rust-best-practices-agent	Fenced JSON {success,data,error} with findings keyed by stable practice id
Architecture or plan review	rust-architect	Fenced JSON {success,data,error} blueprint or design findings
Code review for structural findings	rust-code-reviewer	Fenced JSON {success,data,error} findings summary
Pattern discovery across a codebase	rust-code-explorer	Fenced JSON {success,data,error} with located files and pattern usage

Invoke these agents via Agent Runner using .claude/agents/registry.yaml. Require them to load rust-best-practices and the specific pattern references relevant to the requested review.

Dedicated rust-best-practices-agent assignment template:

{
  "review_mode": "doc_review | sprint_review | phase_end",
  "worktree_path": "/absolute/path/to/worktree",
  "review_targets": [
    "src/",
    "Cargo.toml"
  ],
  "practice_mode": "all | selected",
  "practice_ids": [
    "RBP-001",
    "RBP-004"
  ],
  "notes": "optional context"
}

Output Expectations

When using this skill for review:

• identify the practices reviewed by stable practice id
• report all real findings in scope
• include file:line and remediation guidance for each finding
• treat severity as ordering, not as permission to omit a real finding
• do not dismiss findings as pre-existing or not worsened

Relationship to Other Skills

• rust-development: general Rust style, idioms, documentation, and implementation guidance
• rust-best-practices: structural correctness patterns and lifecycle enforcement
• rust-service-hardening: service-runtime production defaults for Tokio and similar Rust services