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

{
Issue #28599 is harder because it crosses streaming error classification, mid-stream recovery, router fallback behavior, and regression testing across provider-specific paths. Issue #35334 is comparatively localized to request-parameter sanitization in one translation path with a focused regression test.
}
https://github.com/berriai/litellm/issues/28599 4:1 https://github.com/berriai/litellm/issues/35334