A practical walkthrough of Google's Gemini CLI from a DevOps/SRE perspective, covering installation, core commands, headless mode, and key AI concepts. Demonstrates how to use agents, sub-agents, skills (on-demand context loading), plan mode, and MCP (Model Context Protocol) with a Kubernetes example. Skills allow splitting instructions into files loaded only when needed, preventing context window pollution. The guide also shows how deterministic shell scripts within skills prevent LLM guessing errors, and how MCP servers enable natural language interaction with infrastructure like Kubernetes clusters.
•37m watch time
1 Impression