What problem does it solve? When running a multi-service LLM medical consultation system, failures can occur at any layer—Docker MySQL, Ollama, the Python FastAPI server, or Spring Boot—and it is hard to know where to start. This Skill classifies error messages, checks infrastructure health, and applies known fixes for each failure category. ## Core Features & Use Cases - Error Classification: Maps error keywords (e.g., Connection refused :3307, Connection refused :11434) to the failing component such as MySQL, Ollama, the Python LLM server, or Spring Boot. - Infrastructure Health Checks: Provides concrete commands like docker-compose ps, curl http://localhost:11434/api/tags, and curl http://localhost:8000/health to verify each service. - Category-Specific Fixes: Covers common issues including MySQL connection failures, missing Ollama models, Python dependency problems, Spring Boot to Python connectivity (503 errors), garbled Korean/Chinese LLM output, and JPA schema mismatches. - Use Case: You start the project and get "LLM 서버를 사용할 수 없습니다" (503). The Skill walks you through testing the Python endpoint directly, checking the startup order (MySQL → Ollama → Python → Spring Boot), and verifying llm.service.url in application.yml. ## Quick Start Type /troubleshoot or paste your error message and ask the assistant to diagnose why the server is not starting.
