INNER CODE UNIT · Python

main

wgpsec/AboutSecurity · scripts/bench-skill.py:528

def main():
    parser = argparse.ArgumentParser(
        description="Benchmark skill quality with A/B testing (with_skill vs baseline)"
    )
    parser.add_argument(
        "--skill", type=Path,
        help="Path to skill directory (relative to repo root or absolute)"
    )
    parser.add_argument("--all", action="store_true", help="Benchmark all skills with evals")
    parser.add_argument("--runs", type=int, default=1, help="Runs per configuration (default: 1)")
    parser.add_argument("--model", default=None, help="Model for claude -p")
    parser.add_argument("--timeout", type=int, default=120, help="Timeout per run in seconds")
    parser.add_argument("--grade-only", action="store_true", help="Skip running, just re-grade")
    parser.add_argument("--quiet", action="store_true", help="Minimal output")

    args = parser.parse_args()

    if args.all:

View source record →