skills.rest
drunkcoding
drunkcodingCOMMUNITY
Explore all skills by this author →
WebsiteGitHub

generate-profile

Profile an SGLang server end-to-end and generate Chrome-compatible trace files.

4|2|Updated Feb 19, 2026
Search Tags:#performance#profiling#trace#server#cuda#sglang
One-click install
npx skills add https://github.com/drunkcoding/AgentSkillsArxiv --skill generate-profile-drunkcoding
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: generate-profile
Source: https://github.com/drunkcoding/AgentSkillsArxiv/tree/main/skills/sglang/generate-profile
Command: npx skills add https://github.com/drunkcoding/AgentSkillsArxiv --skill generate-profile-drunkcoding

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

This skill profiles an SGLang server end-to-end to generate Chrome-compatible profiling traces, enabling performance diagnosis and validation of server readiness.

Core Features & Use Cases

  • Launches an SGLang server with a given model and port
  • Validates server readiness and basic accuracy before profiling
  • Generates a Chrome-compatible trace directory containing trace files and configuration
  • Returns the profile path for downstream analysis and reproducibility

Quick Start

Launch the SGLang server, run the profiling sequence, and retrieve the generated trace directory.

Frequently Asked Questions about generate-profile

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

FAQPage Schema
How do I profile an SGLang server to generate Chrome-compatible traces?▼

To profile an SGLang server, launch the server with your model and port, validate readiness, and run the profiling sequence to generate a Chrome-compatible trace directory for performance diagnostics.

What is end-to-end performance profiling for CUDA GPUs used for?▼

End-to-end performance profiling for CUDA GPUs is used for latency profiling and diagnosing server readiness, generating trace files and configuration outputs to validate deployment scenarios.

Do I need a CUDA-enabled GPU to capture SGLang profiling traces?▼

Yes, capturing SGLang profiling traces requires a CUDA-enabled GPU and a working SGLang installation to execute the profiling sequence and output the trace directory.

Can I validate SGLang server accuracy before generating a profiling trace?▼

Yes, the profiling process validates SGLang server readiness and basic accuracy before generating the Chrome-compatible trace files and server_args.json configuration.

What is the best way to diagnose latency issues on a deployed SGLang server?▼

The best way to diagnose latency issues is to run an end-to-end profiling trace on the SGLang server, which returns a trace directory containing files and configuration for downstream analysis.

What files do I get from an SGLang profiling trace?▼

An SGLang profiling trace outputs a trace directory containing Chrome-compatible trace files and a server_args.json configuration file to ensure reproducibility for downstream analysis.

Related Skills

Explore More Software Engineering Skills→
FutureMLS-LabFutureMLS-Lab
556

generate-profile

Generate end-to-end performance profiling traces for SGLang server runs.

Official
Intermediate
annealing-inversionannealing-inversion

generate-profile

Profile an SGLang server run and generate a Chrome trace and server_args.json.

Community
Advanced
Nabilhassan12345Nabilhassan12345

generate-profile

Launch an SGLang server and generate a Chrome profiling trace.

Community
Advanced
involvexinvolvex
1

browser-debugger

Debug frontend code in the browser with breakpoints and console inspection.

Community
Advanced