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

{
The right issue is harder because it requires redesigning failure-safe queue and persistence semantics, coordinating retries, locking, recovery, and regression coverage. The left issue is comparatively localized to input normalization and provider-specific translation with focused tests.
}
https://github.com/berriai/litellm/issues/33873 4:1 https://github.com/berriai/litellm/issues/27276