benchmark

Automate vLLM serving benchmarks and generate PR-ready markdown reports.

1|Updated Apr 9, 2026
One-click install
npx skills add https://github.com/WindChimeRan/claude-dotfiles --skill benchmark-windchimeran
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: benchmark
Source: https://github.com/WindChimeRan/claude-dotfiles/tree/main/skills/benchmark
Command: npx skills add https://github.com/WindChimeRan/claude-dotfiles --skill benchmark-windchimeran

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

This Skill automates benchmarking of vLLM serving for the vllm-metal project, including server startup, client benchmarking, branch comparisons, and PR-ready reports.

Core Features & Use Cases

  • Server orchestration and health checks during benchmark runs.
  • Branch/config comparison to identify performance regressions.
  • Auto-generation of PR-ready markdown blocks and matplotlib figures for easy review.

Quick Start

Run the benchmark workflow to compare the current branch against main or other configurations and generate PR-ready results.

Frequently Asked Questions about benchmark

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I automate vLLM serving benchmark tests across different git branches?

You can automate vLLM serving benchmark tests by orchestrating server startup, executing client benchmarks, parsing logs, and generating markdown reports to compare performance across git branches and configurations.

How do I generate PR-ready performance reports for vLLM configuration comparisons?

To generate PR-ready performance reports, the benchmarking process parses client benchmark logs and automatically produces markdown blocks with GitHub formatting and matplotlib figures for immediate pull request review.

Can I use this benchmarking workflow to detect performance regressions in vLLM serving?

Yes, you can use this benchmarking workflow to detect performance regressions by comparing current branch metrics against main or other configurations, ensuring vLLM serving efficiency is validated during PR reviews.

Does the vLLM serving benchmark handle server orchestration and health checks?

Yes, the vLLM serving benchmark handles server orchestration and performs health checks during the benchmark runs to ensure the environment remains stable while executing client benchmarks and parsing logs.

What is the best way to compare vLLM serving performance between configurations?

The best way to compare vLLM serving performance is to run an automated benchmark workflow that starts the server, executes client benchmarks, and generates PR-ready markdown with matplotlib figures for clear visual comparison.