Observe an agent run and GroundEval drafts the contract and diagram for you, no hand-written policy required, then scores what it checked, what it skipped, and what it wasn't allowed to touch.
benchmarking evaluation-framework ai-agents test-agents llm-evaluation llm-as-judge llm-as-a-judge agent-evals agent-evaluation agent-harness agent-loop agent-eval debug-agents
-
Updated
Jul 6, 2026 - Python