Polishing

[johnpmitsch]

Note

Medium Risk
Touches error propagation and exception surfaces across Rust, Python, Node, and Ruby; mistakes could change how failures are classified or break consumer error-handling expectations, though changes are localized to mapping/wrapper layers and examples.

Overview
Introduces a typed, cross-language error hierarchy for the SDK, expanding core SdkError handling with UrlParse/Config surfacing and a new HttpKind classifier (Timeout/Connect/Other) to consistently categorize transport failures.

Bindings are updated to map SdkError into language-native subclasses (Python/Ruby exception classes with status/body metadata; Node via tagged napi::Error messages parsed and rethrown by a new JS wrapper + Proxy that wraps all client methods). Documentation and examples across Rust/Python/Node/Ruby are updated to demonstrate catching specific errors, and small e2e/example tweaks adjust rate-limit/test timing behavior.

^{Reviewed by Cursor Bugbot for commit 11adbb4. Bugbot is set up for automated code reviews on this repo. Configure here.}

[cursor]

Cursor Bugbot has reviewed your changes and found 1 potential issue.

^{❌ Bugbot Autofix is OFF. To automatically fix reported issues with cloud agents, enable autofix in the Cursor dashboard.}

^{Reviewed by Cursor Bugbot for commit fb4e06a. Configure here.}

[cursor]

Trailing \x1f control character in Node.js error messages

Medium Severity

The Rust format! on line 34 unconditionally appends \u{001f} (unit separator) even when body_s is empty (for Config, Http, Timeout, Connect variants). On the JS side, fromNapiError only strips the separator when bodyLenStr !== "-", so non-body error variants end up with a trailing \x1f control character in their .message. This pollutes user-facing error messages and can break string comparisons, JSON serialization, and log output.

Additional Locations (1)

• npm/errors.js#L66-L77

 

^{Reviewed by Cursor Bugbot for commit fb4e06a. Configure here.}