/benchmark $ARGUMENTS
Run a repeatable benchmark command on the current branch and compare it with a baseline ref. This skill is for measurement, not speculation.
If the user wants architectural performance review without running code, use /multi-review with performance-oracle instead.
Arguments
/benchmark "<command>" [--baseline <ref>] [--iterations N]
Defaults:
--baseline origin/main--iterations 5- warmup runs:
1
Reject the request if no command is provid
[Description truncada. Veja o README completo no GitHub.]