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

{
The authentication regression is harder because it likely requires tracing proxy token lookup, caching, database compatibility, upgrade behavior, and multiple request paths, with higher regression risk. The media feature is comparatively bounded to provider-specific content normalization, validation, and targeted multimodal tests.
}
https://github.com/berriai/litellm/issues/24680 3:2 https://github.com/berriai/litellm/issues/30501