Can LLM-as-a-Judge Reliably Verify Rubrics in Agentic Scenarios?

[system]

---

This is a companion discussion topic for the original entry at https://arxiv.org/abs/2606.29920