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

{
The guardrail defect is harder because it requires safely reconciling verdict and redaction semantics, defining fail-open versus fail-closed behavior, and adding regression coverage for malformed or incomplete detector output. The Azure model addition is comparatively scoped to provider integration, capability metadata, and compatibility tests.
}
https://github.com/berriai/litellm/issues/30730 3:1 https://github.com/berriai/litellm/issues/32613