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

{
The right issue is harder because it involves defensive handling across an observability callback, compatibility with multiple response shapes, validation of attribute types, and regression testing across integration paths. The left issue appears localized to parameter propagation at a single health-check call site.
}
https://github.com/berriai/litellm/issues/24516 3:1 https://github.com/berriai/litellm/issues/29266