Agent Instructions

This repository is a Rust engineering skill system for AI coding agents.

Quick start

1. Load the root router (SKILL.md) for any Rust task.
2. Route to the appropriate skill based on the task.
3. Load references when deep dives are needed.
4. Consult examples for compile-tested patterns.

Skill catalog

Native core

• rust-fundamentals — language basics
• rust-std — standard library API discovery
• rust-production — production engineering
• rust-debugging — debugging
• rust-unsafe-checker — unsafe Rust review
• rust-dependencies — dependency management

Security

• rust-code-audit — repository security audit

Library extensions

• rust-tokio — async runtime
• rust-axum — HTTP
• rust-serde — serialization
• rust-diesel — databases
• rust-rayon — parallelism
• rust-clap — CLI
• rust-ratatui — TUI

Principles

• Native Rust first. Prefer std over third-party crates.
• No blocking in async. Use spawn_blocking or async-native crates.
• No casual unwrap() in production paths.
• Version-sensitive advice names its version.
• Examples use edition 2021.

Agent execution safety

• Read-only actions (inspect, search, analyze, explain, review) are always safe.
• Mutating actions (create, modify, delete, install, commit, push, execute) require user awareness.
• Destructive operations require explicit user confirmation.
• Always: analyze first, plan second, modify third, validate fourth.

Validation

bash scripts/validate.sh
bash scripts/test-triggers.sh