---
name: Perf Benchmarker
slug: perf-benchmarker
category: Quality
description: Perf Benchmarker enforces sequential performance benchmarks with minimum durations and a JSON metrics output contract. Use it when establishing baselines or validating regressions without parallel runs.
github: "https://github.com/composio-community/awesome-claude-plugins/tree/master/perf/skills/benchmark"
language: JavaScript
stars: 1892
forks: 565
install: "npx degit https://github.com/composio-community/awesome-claude-plugins/tree/master/perf/skills/benchmark ~/.claude/skills/benchmark"
installs_to: ~/.claude/skills/benchmark
source_path: perf/skills/benchmark/SKILL.md
collection_size: 23
category_size: 1354
collection_url: "https://dirskills.com/collections/composio-community/awesome-claude-plugins"
added: 2026-08-18T06:59:17.998Z
last_synced: 2026-08-18T06:59:17.998Z
canonical_url: "https://dirskills.com/skills/perf-benchmarker"
---

# Perf Benchmarker

Perf Benchmarker enforces sequential performance benchmarks with minimum durations and a JSON metrics output contract. Use it when establishing baselines or validating regressions without parallel runs.

**Install:**

```bash
npx degit https://github.com/composio-community/awesome-claude-plugins/tree/master/perf/skills/benchmark ~/.claude/skills/benchmark
```

## README

# perf-benchmarker

Run sequential benchmarks with strict duration rules.

Follow `docs/perf-requirements.md` as the canonical contract.

## Required Rules

- Benchmarks MUST run sequentially (never parallel).
- Minimum duration: 60s per run (30s only for binary search).
- Warmup: 10s minimum before measurement.
- Re-run anomalies.

## Output Format

```
command: <benchmark command>
duration: <seconds>
warmup: <seconds>
results: <metrics summary>
notes: <anomalies or reruns>
```

## Output Contract

Benchmarks MUST emit a JSON metrics block between markers:

```
PERF_METRICS_START
{"scenarios":{"low":{"latency_ms":120},"high":{"latency_ms":450}}}
PERF_METRICS_END
```

## Constraints

- No short runs unless binary-search phase.
- Do not change code while benchmarking.