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

{
#35068 is harder because it spans proxy error handling, route classification, metadata precedence, retry behavior, and spend-log attribution, requiring cross-cutting investigation and regression coverage. #33184 is comparatively localized to parameter propagation through the OpenAI translation path with focused adapter and proxy tests.
}
https://github.com/berriai/litellm/issues/35068 3:1 https://github.com/berriai/litellm/issues/33184