skills.rest
annealing-inversion
annealing-inversionCOMMUNITY
Explore all skills by this author →
GitHub

generate-profile

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

Updated Jun 4, 2026
Search Tags:#performance#profiling#benchmark#profile#sglang#server-trace
One-click install
npx skills add https://github.com/annealing-inversion/sglang-kimi-deferral --skill generate-profile-annealing-inversion
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: generate-profile
Source: https://github.com/annealing-inversion/sglang-kimi-deferral/tree/main/.claude/skills/generate-profile
Command: npx skills add https://github.com/annealing-inversion/sglang-kimi-deferral --skill generate-profile-annealing-inversion

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Automates the end-to-end profiling of an SGLang server run, producing a Chrome-compatible trace and a ready-to-review server configuration to accelerate performance optimization.

Core Features & Use Cases

  • Launch the SGLang server with a given model and port, then monitor readiness via the health endpoint to ensure deterministic profiling.
  • Validate server accuracy with a quick evaluation step to ensure meaningful profiling data before tracing.
  • Generate profiling traces, producing a timestamped trace directory (Chrome/Perfetto compatible) and a server_args.json file for reproducibility.

Quick Start

Run the profile workflow against a live SGLang server to generate a profiling trace.

Frequently Asked Questions about generate-profile

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

FAQPage Schema
How do I capture an end-to-end SGLang profiling trace?▼

To capture an SGLang profiling trace, you automate the server run to generate a Chrome-compatible trace and a server_args.json file. This workflow monitors server readiness and validates accuracy before collecting performance data.

What is the best way to benchmark SGLang startup latency and performance?▼

Benchmarking SGLang startup latency is best done by automating end-to-end profiling to produce a timestamped trace directory. This approach ensures deterministic profiling by monitoring the health endpoint before data collection.

Do I need a CUDA-enabled GPU to generate an SGLang server trace?▼

Yes, generating an SGLang server trace requires a CUDA-enabled GPU. You also need a working SGLang installation to execute the profiling workflow and output the reproducible server configuration.

How does server validation work during SGLang profiling?▼

Server validation during SGLang profiling works by monitoring the health endpoint for readiness and running a quick evaluation step. This ensures meaningful and accurate profiling data is collected before tracing begins.

Can I view SGLang profiling traces in Chrome or Perfetto?▼

Yes, you can view SGLang profiling traces in Chrome or Perfetto. The profiling workflow outputs a timestamped trace directory specifically formatted to be compatible with these visualization tools.

Related Skills

Explore More Software Engineering Skills→
drunkcodingdrunkcoding
4

generate-profile

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

Community
Advanced
klxiaoniuklxiaoniu
174

perfetto-trace-analysis

Analyze Perfetto trace files to identify Android latency, memory, or UI jank root causes.

Community
Advanced
QingHuan-6QingHuan-6

sglang-diffusion-benchmark-profile

Measure diffusion denoise latency and collect performance traces for SGLang.

Community
Advanced
Fridge003Fridge003
9

generate-profile

Generate a Chrome-compatible profiling trace for an SGLang server run.

Community
Advanced