automated-evaluation
Autonomous failure investigation, root-cause analysis, and validated interventions for LLMs, VLMs, and agents.
Open protocol for testing whether ML and AI-safety metrics add information beyond cheap baselines.
