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

{
17696 requires provider-aware request transformation, token-threshold handling, and careful interaction with retries and caching semantics; 35354 appears comparatively localized to terminal-state bookkeeping with focused regression coverage.
}
https://github.com/berriai/litellm/issues/17696 3:1 https://github.com/berriai/litellm/issues/35354