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

{
#32068 requires cross-layer work across proxy streaming, cache serialization/deserialization, replay semantics, and regression coverage, while #32900 is comparatively localized to wizard/provider configuration and validation.
}
https://github.com/berriai/litellm/issues/32068 4:1 https://github.com/berriai/litellm/issues/32900