OpenCastor Agent Harness Evaluator Leaderboard
263k config search space benchmarked across robot fleets—nothing like this exists for robotics AI.
An Open Benchmark for AI in Cybersecurity Operations.
Ablation testing isolates tool vs playbook contribution in AI security evals.
Security teams, AI researchers, SOC analysts
MLCommons · HELM · Security AI benchmarks
263k config search space benchmarked across robot fleets—nothing like this exists for robotics AI.
Local-only memory benchmarking—rigorous, but LoCoMo is a standard test everyone runs.
Deterministic agent benchmarking with strict validation—unlike SWE-Bench, measures whether agents actually operate.
Mining your own bugfix history for evals beats public benchmarks that don't match your codebase.
Unsupervised bug benchmark using agents as both attackers and defenders—novel scoring methodology.
Research article revealing few-shot collapse patterns, not a usable tool or product.