skills.rest
QingHuan-6
QingHuan-6COMMUNITY
Explore all skills by this author →
GitHub

sglang-prod-incident-triage

Convert live SGLang serving symptoms into reproducible replays and structured next-step guidance.

Updated May 30, 2026
Search Tags:#debugging#production#triage#replay#sglang#incident-triage
One-click install
npx skills add https://github.com/QingHuan-6/SglangGen --skill sglang-prod-incident-triage-qinghuan-6
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: sglang-prod-incident-triage
Source: https://github.com/QingHuan-6/SglangGen/tree/main/.claude/skills/sglang-prod-incident-triage
Command: npx skills add https://github.com/QingHuan-6/SglangGen --skill sglang-prod-incident-triage-qinghuan-6

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill requires requests, and includes scripts (resource) and references (resource) components.

What problem does it solve?

Replay-first debug flow for SGLang serving problems. Use this skill to turn a live or recent server issue—such as health-check failures, latency or throughput regressions, queue growth, timeouts, distributed stalls, crash dumps, or wrong outputs after a deploy—into a replayable path and the right next debugging tool.

Core Features & Use Cases

  • Collect a baseline bundle, capture the failing request or crash dump, and replay on a clean target to isolate the problem.
  • Leverage focused tools (debug-cuda-crash, debug-distributed-hang, llm-torch-profiler-analysis) when the boundary between components is unclear.
  • Use established references and case studies to guide triage and decision making under pressure.

Quick Start

Collect a read-only live bundle, save the failing request or crash dump, and replay it on a clean target to begin the triage.

Frequently Asked Questions about sglang-prod-incident-triage

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

FAQPage Schema
How do I triage SGLang production incidents like latency regressions or queue growth?▼

You can triage distributed stalls by capturing a live bundle and replaying the failing request on a clean target to isolate the issue. This replay-first flow provides structured next-step guidance for SGLang distributed stalls.

What is the best way to debug SGLang health-check failures after a deploy?▼

The best way to debug SGLang health-check failures is collecting a baseline bundle and replaying the failing request on a clean target. This isolates post-deploy output changes into structured next-step guidance.

Can I use this triage flow for SGLang PD, EP, and HiCache configurations?▼

Yes, this triage flow supports SGLang PD, EP, and HiCache configurations. It handles health-check failures, latency regressions, and crash dumps across these distributed serving setups by generating reproducible replays.

When should I switch to focused debugging tools during an SGLang incident triage?▼

Switch to focused debugging tools like debug-cuda-crash or debug-distributed-hang when the boundary between SGLang components is unclear. The triage flow leverages these tools to provide clear next steps under pressure.

Do I need to install specific dependencies to replay SGLang server crashes?▼

You need the requests dependency installed to use this SGLang triage flow. It enables the replay-first debug process for capturing live bundles and converting SGLang server crashes into reproducible debug paths.

Related Skills

Explore More Software Engineering Skills→
annealing-inversionannealing-inversion

sglang-prod-incident-triage

Replay live SGLang serving faults into reproducible debug paths.

Community
Advanced
dyyounggdyyoungg

sglang-prod-incident-triage

Turn live SGLang serving incidents into replay-based debugging workflows.

Community
Advanced
Nabilhassan12345Nabilhassan12345

debug-cuda-crash

Diagnose CUDA crashes in SGLang with kernel API logging.

Community
Advanced
hey-konghey-kong

sglang-prod-incident-triage

Replay failing SGLang requests on a clean target for incident triage.

Community
Advanced