Day 1
Understand
How agents work and how to steer them
- LLM and agent fundamentalsTokens, reasoning, why models are stateless and answer differently; model choice, benchmarks vs. reality
- The Claude Code toolkitCLAUDE.md, rules, skills, commands, subagents, hooks, output styles, MCP, plugins: which tool for which job
- Prompting and context engineeringStructuring prompts, curating context, context rot, hallucinations, avoiding overload