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

{
#32574 is harder because it requires tracing and safely changing shared logging state across router retries, fallbacks, callbacks, and cooldown behavior, with broader regression and concurrency risk. #27532 is comparatively localized to request-header inference and provider-specific validation.
}
https://github.com/berriai/litellm/issues/32574 3:1 https://github.com/berriai/litellm/issues/27532