Feature request: Add Giskard-based evaluation integration for regression testing of LiveKit agents

Agents
1

Hi maintainers, I’m planning a contribution and want to confirm scope before
opening a GitHub issue.

Proposal:

  • Add Giskard integration for LiveKit Agents focused on eval/regression
    testing (not runtime STT/TTS/LLM provider integration).
  • Goal: run repeatable quality checks and detect regressions across prompt/
    model/plugin changes.

Questions:

  1. Does this direction fit LiveKit’s roadmap?
  2. Preferred location:
    • core evals API in livekit-agents/livekit/agents/evals
    • or a separate package under livekit-plugins/livekit-plugins-giskard
  3. For MVP, would maintainers accept:
    • adapter + example + tests
    • CI-friendly output for pass/fail regression checks

If this sounds good, I’ll open a feature request issue with concrete API and
acceptance criteria.