Auto-generated agent evaluation benchmark with 1,040 tasks across 104 unique scenarios created by ClawEnvKit.

	Statistics




	




	

Tasks 1,040

Categories 24

Mock services 20

Task types API-based (77%) + file-dependent (23%)

huggingface-cli download AIcell/Auto-ClawEval --repo-type dataset --local-dir Auto-ClawEval

bash run_harnesses.sh --harness claudecode… See the full description on the dataset page: https://huggingface.co/datasets/AIcell/Auto-ClawEval.