retryx

I built retryx because most retry libraries in Rust either come with too much policy surface area or hide behavior behind macros. I wanted something tiny, explicit, and async-first—where you can see and control every retry, delay, and timeout in one fluent builder call.

retryx wraps any async operation that returns Result<T, E> and retries it with fixed or exponential backoff. It includes per-attempt timeouts, custom retry filters, and an on_retry hook for logging or metrics—all without macros or unsafe code.

What Makes It Different

• Async-first design built for Tokio and async/await
• No macros, no unsafe—simple, readable implementation
• No global state, no background tasks—runs in caller's async context
• Fluent builder API—configure retries, delays, timeouts, and filters in one chain
• Custom retry filters—decide which errors are retryable
• Per-attempt timeouts—wraps each attempt in tokio::time::timeout

Built & Shipped

Pure Rust crate with zero dependencies beyond Tokio. Guarantees no panics, no global state, no background tasks. Available on crates.io with full documentation and examples.