benchmark

You can automate benchmarking by executing a user-specified command to collect latency and throughput metrics, storing results in an Obsidian vault, and comparing current runs against a saved baseline to flag regressions beyond a defined threshold.

To store benchmark results in an Obsidian vault, the Skill captures stdout and stderr from your benchmark command, extracts numeric performance metrics, and saves them in a structured report format for consistent auditing over time.

Yes, the Skill executes arbitrary user-provided benchmark commands and extracts numeric metrics like latency, throughput, and resource usage, applying to any project requiring consistent performance evaluation.

The best way to compare benchmark results is to save an initial performance run as a baseline, then automatically compare subsequent command outputs against it to generate regression alerts for quick remediation.

Regression detection works by comparing newly collected performance metrics against a stored baseline, automatically flagging any latency or throughput degradation that exceeds your pre-defined threshold in a structured report.

No specific framework is needed, as the Skill applies to any project by executing your custom benchmark commands and capturing standard output and error streams to extract structured performance data.