1. src/providers/yubikey_usb.rs, line 398-401 (link)

   write_packet error message conflates libusb error codes with short-write counts

   control_transfer returns either a non-negative byte count on success or a negative libusb error code (e.g., -3 = LIBUSB_ERROR_ACCESS, -6 = LIBUSB_ERROR_BUSY) on failure. The current condition rc != 8 correctly catches both cases, but the message "wrote -3, expected 8" reads as if the device returned a 3-byte payload rather than surfacing the underlying error code. This makes it harder to diagnose access/permission issues on Linux (e.g., udev rules missing) vs. genuine HID write failures.

   Distinguishing the two cases makes the error actionable:

2. src/providers/yubikey_usb.rs, line 496-505 (link)

   release_interface called unconditionally even when interface was never claimed

   find_and_open passes through the claim_interface call returning LIBUSB_ERROR_NOT_SUPPORTED (-12) without error, which on macOS means the IOKit HID subsystem owns the device and no interface was claimed. DeviceHandle::drop then calls release_interface on an interface that was never claimed, which libusb returns an error for (silently ignored here). While harmless in practice, it triggers a spurious kernel error log entry on some versions of macOS and will confuse future readers auditing the cleanup path.

   The standard fix is to mirror the driver_was_detached pattern with an interface_claimed field:

   struct DeviceHandle<'a> {
       lib: &'a LibUsb,
       handle: *mut LibusbDeviceHandle,
       driver_was_detached: bool,
       interface_claimed: bool,   // true when claim_interface returned 0
   }

   Then in drop:

   if self.interface_claimed {
       (self.lib.release_interface)(self.handle, 0);
   }
   if self.driver_was_detached {
       (self.lib.attach_kernel_driver)(self.handle, 0);
   }
   (self.lib.close)(self.handle);

3. src/providers/yubikey_usb.rs, line 452-492 (link)

   read_response return value does not represent bytes read

   The function signature -> Result<usize> implies a byte count, but the returned value is r0 — the start offset of the last written chunk, not the number of response bytes in the buffer. For a 20-byte HMAC + 2-byte CRC response, r0 ends up at 21 (the position of the fourth chunk's start) rather than 22 (the bytes actually populated). The caller in challenge_response_hmac discards this value (handle.read_response(&mut response)?;), so there is no current bug, but the signature is a future trap.

   Consider returning a meaningful constant (22 for the data + CRC length) or changing the return type to Result<()> if the count is not needed.

• Cargo.lock
  • Removed numerous transitive dependencies associated with yubico_manager, rusb, and libusb1-sys.
  • Added libloading as a new dependency.
  • Generalized many other dependency version specifications.
• Cargo.toml
  • Replaced the yubico_manager dependency with libloading.
• src/providers/mod.rs
  • Added a new module declaration for yubikey_usb.
• src/providers/yubikey.rs
  • Refactored YubiKey operations to use the new yubikey_usb module instead of yubico_manager.
• src/providers/yubikey_usb.rs
  • Introduced a new module containing the custom implementation for dynamic libusb loading, device enumeration, control transfers, and YubiKey HMAC-SHA1 challenge-response logic.

• The pull request summary and initial commit were generated by "Claude Code" and "Cursor Bugbot", indicating automated assistance in its creation.

The full guide for Gemini Code Assist can be found on our documentation page, here are some quick tips.

Invoking Gemini

You can request assistance from Gemini at any point by creating a comment using either /gemini <command> or @gemini-code-assist <command>. Below is a summary of the supported commands on the current page.

Customization

To customize Gemini Code Assist for GitHub experience, repository maintainers can create a configuration file and/or provide a custom code review style guide (such as PEP-8 for Python) by creating and adding files to a .gemini/ folder in the base of the repository. Detailed instructions can be found here.

Limitations & Feedback

Gemini Code Assist may make mistakes. Please leave feedback on any instances where its feedback is incorrect or counter productive. You can react with 👍 and 👎 on @gemini-code-assist comments. If you're interested in giving your feedback about your experience with Gemini Code Assist for GitHub and other Google products, sign up here.

You can also get AI-powered code generation, chat, as well as code reviews directly in the IDE at no cost with the Gemini Code Assist IDE Extension.