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

{
The left issue requires cross-layer streaming behavior changes, careful protocol compatibility, timing-sensitive async handling, and end-to-end coverage across deployment configurations. The right issue is a comparatively localized lifecycle fix involving task cancellation and regression tests, with narrower behavioral risk.
}
https://github.com/berriai/litellm/issues/34819 4:1 https://github.com/berriai/litellm/issues/31857