Sorting LiteLLM GitHub issues by difficulty to implement. Higher = harder.

{
The left issue requires debugging stateful realtime event sequencing, coordinating proxy behavior with backend VAD semantics, and validating timing-sensitive regressions across streaming scenarios. The right issue is comparatively localized to response schema propagation and focused unit or integration coverage.
}
https://github.com/berriai/litellm/issues/31726 4:1 https://github.com/berriai/litellm/issues/33967