Agentic AI Learning Portal

Appearance

Day 2 -- Gemini CLI: Advanced Features ​

Day 1 got Gemini CLI running and taught you the essentials: installation, authentication, the REPL, and basic coding tasks. Now we go deeper. Day 2 covers the features that turn Gemini CLI from a terminal assistant into a platform you can extend, secure, and integrate into real development workflows. By the end of today you will have MCP servers connecting Gemini CLI to external tools, custom slash commands that encode your team's conventions, extensions that bundle everything into shareable packages, sandboxing that keeps experimental code off your machine, and an understanding of which release channel fits your risk tolerance.

What you will have by the end of today ​

  • One or more MCP servers connected to Gemini CLI, with a working understanding of transport types, trust settings, and tool allowlists
  • At least one custom slash command defined as a TOML file, with argument substitution and shell command expansion
  • An extension installed from the gallery, with awareness of how extensions package commands, MCP servers, hooks, and skills together
  • Sandboxing configured for your platform, with the confidence to run untrusted code safely
  • A release channel selected that matches your stability needs

Before you start ​

ToolVersionPurposeInstall
Gemini CLIv0.45+Terminal agent (all lessons)npm install -g @google/gemini-cli
Docker24+Container sandbox (sandbox lesson)docker.com
Python3.10+MCP server examplespython.org
bash
# Ensure you are on a recent version
npm install -g @google/gemini-cli@latest
gemini --version

All MCP and slash command features covered here require Gemini CLI v0.45 or later. If you installed in Day 1 and have not updated since, run the command above before starting.

Lessons ​

  1. MCP Servers -- connecting Gemini CLI to external tools and data sources through the Model Context Protocol, configuring transports, managing trust, and understanding how MCP Prompts become slash commands
  2. Custom Slash Commands -- defining reusable prompts as TOML files, argument substitution with , shell command expansion, namespacing, and scoping commands to users or projects
  3. Extensions -- installing and managing extensions from the gallery, understanding what extensions package, building custom tool integrations, and the partner ecosystem
  4. Release Channels -- choosing between stable, preview, and nightly releases, understanding release cadence, and picking the right channel for your workflow
  5. Gemini Code Assist -- Google's IDE companion for VS Code, JetBrains, Cloud Workstations, and Cloud Shell Editor, with code completion, generation, and chat

Next: Day 3 -- Antigravity: IDE & Platform