1. Explain what an LLM is and how it relates to AI more broadly
  2. Describe, in plain terms, how an LLM generates a response: training data, tokens, and predicting what comes next, and why that mechanism can produce confidently wrong answers
  3. Explain why context and caching matter: how conversation and session context shape an LLM's answers, and how caching lets a tool reuse that context efficiently instead of reprocessing it from scratch every time
  4. Recognize that different companies build different LLMs, and why they vary in capability, speed, and cost
  5. Describe, in plain terms, what skills, plugins, and MCP (Model Context Protocol) connections are, and how they let an AI tool go beyond plain text generation to use external tools, data, or context
  6. Recognize data security and privacy considerations when sharing code, data, or business information with an AI tool