INNER CODE UNIT · Python
aggregate_benchmark
AgentSkillOS/SkillAnything · scripts/aggregate_benchmark.py:57
def aggregate_benchmark(
workspace_dir: Path | None = None,
run_dirs: list[Path] | None = None,
) -> dict[str, Any]:
"""Aggregate benchmark results from multiple runs.
Args:
workspace_dir: Directory containing run_* subdirectories.
run_dirs: Explicit list of run directories.
Returns:
Dict with aggregated statistics.
"""
if run_dirs is None and workspace_dir is not None:
# Discover run directories
run_dirs = sorted(
d for d in workspace_dir.iterdir()
if d.is_dir() and d.name.startswith("run_")