skills.rest
saintgo7
saintgo7COMMUNITY
Explore all skills by this author →
GitHub

gem-llm-troubleshooting

Map GIL-LLM production symptoms to 13 known failure cases and apply fixes.

Updated Apr 4, 2026
Search Tags:#troubleshooting#incident response#vllm#database recovery#gem-llm#ssh connectivity#oom mitigation
One-click install
npx skills add https://github.com/saintgo7/claude-skills --skill gem-llm-troubleshooting
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: gem-llm-troubleshooting
Source: https://github.com/saintgo7/claude-skills/tree/main/gem-llm-troubleshooting
Command: npx skills add https://github.com/saintgo7/claude-skills --skill gem-llm-troubleshooting

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

It solves the problem of getting stuck during GEM-LLM operations when common runtime failures occur, such as GPU OOM, vLLM model/buffering issues, SSH/connectivity failures, database bottlenecks, and high-level “it’s broken” symptoms.

Core Features & Use Cases

  • Rapid symptom-to-case matching: Quickly map observed symptoms (like 500s, OOM, disk I/O errors, QueuePool warnings, missing model keys, slow responses) to one of 13 known failure cases from GEM-LLM operations.
  • Actionable, case-specific remedies: Provide concrete fixes per case, including environment dependency installs, prompt-routing adjustments, streaming option configuration, and database pool/recovery steps.
  • Operational triage workflow: Establish a short diagnostic flow (supervisor status → logs → keyword mapping) to reduce time-to-resolution during live incidents.

Quick Start

Run install.sh with the gem-llm-troubleshooting skill to load the 13-case troubleshooting reference and follow the symptom-to-case matching workflow.

Frequently Asked Questions about gem-llm-troubleshooting

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

FAQPage Schema
How do I troubleshoot vLLM startup failures and GPU OOM errors during production inference?▼

Troubleshoot vLLM startup failures and GPU out-of-memory errors by mapping observed symptoms to known incident cases. This Skill uses a diagnostic flow checking supervisor status and logs, then matches keywords like missing model keys to specific remediation steps.

What is the best way to fix SSH connectivity problems and disk I/O errors in production LLM infrastructure?▼

Fix SSH connectivity problems and disk I/O errors by following a deterministic triage workflow. This Skill maps filesystem and connectivity symptoms to a curated set of known incident cases, applying targeted remediation guidance to resolve specific failures.

How do I resolve database QueuePool warnings and corruption events when serving LLM applications?▼

Resolve database QueuePool warnings and corruption events by matching diagnostic keywords to a dedicated case playbook. This Skill provides concrete fixes including database pool adjustments and recovery steps to restore performance during live incidents.

How do I diagnose prompt schema mismatches and streaming issues causing 500 errors in vLLM?▼

Diagnose prompt schema mismatches and streaming issues causing 500 errors by applying a symptom-to-case matching workflow. This Skill checks logs for diagnostic keywords and suggests streaming option configurations and prompt-routing adjustments.

Does this troubleshooting workflow require specific environment dependencies or prior incident response knowledge?▼

This troubleshooting workflow requires no prior incident response knowledge. The Skill establishes a short diagnostic flow from supervisor status to logs, mapping symptoms to 13 known cases, but requires specific environment dependency installs to apply targeted fixes.

Related Skills

Explore More Software Engineering Skills→
vercelvercel
246

investigation-mode

Orchestrate diagnostic triage for stuck or slow applications across logs, workflows, and deployments.

Official
Advanced
calevanscalevans
2

deployment

Automate PHP application deployment to production with Git or rsync.

Community
Advanced
davpatel605-beepdavpatel605-beep

serving-llms-vllm

Deploy and optimize vLLM inference servers with OpenAI-compatible APIs.

Community
Advanced
NightreaverNightreaver
6

ssh-incident-response

Automates diagnostic workflows for SSH host incident response.

Community
Advanced