Select optimal AI models for tasks using GPT-5, GPT-4.1-mini, and Claude MCP.
npx skills add https://github.com/kimny1143/mued_v2 --skill ai-models
Please help me install this Agent Skill. Skill: ai-models Source: https://github.com/kimny1143/mued_v2/tree/main/.claude/skills/ai-models Command: npx skills add https://github.com/kimny1143/mued_v2 --skill ai-models
This Skill helps select appropriate AI models for MUED LMS v2, balancing cost and quality across OpenAI API usage and Claude MCP flow.
For a complex content generation task, use GPT-5; for quick Q&A, use GPT-4.1-mini; for internal testing use Claude via MCP.
High-intent search queries and answers about installing and using this skill.
Select GPT-5 series for complex reasoning tasks requiring deep analysis, GPT-4.1-mini for simple generation and quick Q&A to reduce costs, and Claude via MCP for internal development and testing workflows. Match model complexity to task requirements.
Model selection balances quality and cost by routing complex judgment tasks to GPT-5, simple generation to GPT-4.1-mini, and testing to Claude MCP. This strategy optimizes token usage and aligns spending with task complexity.
Use Claude via MCP during development and internal testing to control costs, then transition to GPT-5 for production complex reasoning or GPT-4.1-mini for production simple generation based on your task type and quality requirements.
Yes. The model selection policy includes Japanese-quality educational content requirements, ensuring language quality aligns with your audience. Select models based on content complexity while maintaining educational standards.
GPT-4o is explicitly excluded from this model selection policy. Use GPT-5 series for complex tasks, GPT-4.1-mini for simple generation, and Claude MCP for testing instead to maintain cost efficiency and alignment with policy constraints.
marcusgollClassify software development task complexity into QUICK, STANDARD, and DEEP tiers.
devendrapratapsinghOptimize AI and LLM operations for token usage, cost, and inference.
ROLLED740Route LLM tasks by complexity to cheaper models with cost tracking.
ZurybrDesign batching schedulers that optimize LLM inference on compilation-based accelerators.
