OpenAI Codex Rust CLI Agent Best Practices

This skill teaches you to write Rust code in the style of the OpenAI Codex codebase - a production CLI/agent system with 50 crates and 787 Rust files.

Key Characteristics

• •Edition 2024 with strict Clippy configuration
• •Zero unwrap/expect in non-test code (enforced at workspace level)
• •Tokio async runtime with proper Send + Sync bounds
• •thiserror for library errors, anyhow for application code
• •Flat workspace structure with centralized dependencies

When to Apply

Apply this skill when:

• •Building CLI tools or agent systems in Rust
• •Writing async Rust with Tokio
• •Designing Rust workspace organization
• •Implementing error handling patterns
• •Working on production Rust codebases

Quick Reference

Critical Rules (Must Follow)

Error Handling

Organization

Component Patterns

Naming Conventions

Style

Cross-Crate

Example: Proper Error Handling

use thiserror::Error;
use anyhow::Context;

// Domain error with thiserror
#[derive(Debug, Error)]
pub enum ConfigError {
    #[error("failed to read config file: {path}")]
    ReadFailed {
        path: PathBuf,
        #[source]
        source: std::io::Error,
    },

    #[error(transparent)]
    Parse(#[from] toml::de::Error),
}

// Library function returns domain error
pub fn load_config(path: &Path) -> Result<Config, ConfigError> {
    let content = fs::read_to_string(path)
        .map_err(|source| ConfigError::ReadFailed {
            path: path.to_owned(),
            source,
        })?;
    toml::from_str(&content).map_err(Into::into)
}

// Application code uses anyhow with context
fn main() -> anyhow::Result<()> {
    let config = load_config(Path::new("config.toml"))
        .context("failed to load configuration")?;
    run(config).await
}

Source

Patterns extracted from OpenAI Codex (codex-rs/ subdirectory) - a production Rust codebase with 50 crates and 787 Rust files.