Measures latency, throughput, size, and memory with reproducible benchmark workflows.
npx skills add https://github.com/nuggocto/dotfiles --skill benchmark-nuggocto
Please help me install this Agent Skill. Skill: benchmark Source: https://github.com/nuggocto/dotfiles/tree/main/opencode/skills/benchmark Command: npx skills add https://github.com/nuggocto/dotfiles --skill benchmark-nuggocto
High-intent search queries and answers about installing and using this skill.
Build production-equivalent artifacts for both baseline and candidate, then run them under identical workloads, hardware, and flags with randomized or interleaved run order. Compare against a predeclared regression threshold and report the effect size with harness-native uncertainty or run-to-run variability.
Common choices include oha, wrk2, vegeta, k6, autocannon, and fortio, depending on whether you need open or closed load models. Report target and achieved arrival rate, errors, and the full latency distribution rather than averages.
Tail percentiles need enough observations beyond the target percentile and cannot be derived from mean and standard deviation. Use raw samples or HDR histograms, report sample counts, and never average per-run percentiles.
No, unless that mode is what ships or is the explicit subject of measurement. Debug builds, hot-reload modes, and build-tool wrapper overhead distort results; always match the production compiler, runtime options, and deployment constraints.
Profiling explains why a result occurred by showing CPU, memory, or allocation hotspots, but it can perturb timing. Use benchmarks to decide whether a change is faster, then profile separately to locate the bottleneck, and validate claims with an unprofiled run.
growilabsGuide test authors to verify observable behavior instead of internal implementation details.
vobbilisEnforce before/after benchmarks for hot-path changes in CI pipelines.
hiromimaProfiles CPU, memory, and benchmarks of Rust applications using cargo, flamegraph, and perf.
az9713Automate web page performance benchmarking with browser automation and JavaScript evaluation.
