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

{
Issue #34399 is harder because it affects router-wide retry and failover behavior across deployment groups, requiring careful handling of shared upstream limits, health state, timing, and regression coverage. Issue #31959 is more localized to request-scoped token allocation and synchronization in the Presidio guardrail flow.
}
https://github.com/berriai/litellm/issues/34399 3:2 https://github.com/berriai/litellm/issues/31959