Skip to main content

Core Commands & Slash Commands

From Setup to Daily Use: Mastering Gemini CLI Commands​

In Lesson 2, you installed Gemini CLI and saw it work for the first time. Now comes the crucial step: learning the commands you'll use every day to work efficiently with your AI development partner.

This isn't about memorizing syntax. It's about understanding when to use each command, why it matters, and how to orchestrate your AI work like a professional.

Understanding Gemini CLI's Three Input Methods​

Gemini CLI has three ways to interact:

1. Slash Commands (/) — Meta-level control

  • Control Gemini CLI behavior
  • Manage conversations and context
  • Configure settings
  • Access integrations

2. At Commands (@) — File/directory inclusion

  • Reference files and folders
  • Include content in prompts
  • Automatic smart filtering

3. Shell Commands (!) — System access

  • Execute bash/cmd commands
  • Interactive shell mode
  • AI-guided execution

Let's dive into each.

Slash Commands: The Complete Reference​

Gemini CLI provides 29+ slash commands. Here are the essential ones you'll use daily:

Conversation Management​

CommandPurposeExample
/chat save <tag>Save current conversation/chat save auth-implementation
/chat resume <tag>Restore saved conversation/chat resume auth-implementation
/chat listList saved conversations/chat list
/chat delete <tag>Remove saved conversation/chat delete auth-implementation
/chat share file.mdExport conversation to file/chat share output.md
/clearClear terminal display/clear

Context & Memory Management​

CommandPurposeExample
/memory add <text>Add to AI's context/memory add we use TypeScript strict mode
/memory showDisplay loaded context/memory show
/memory refreshReload GEMINI.md files/memory refresh
/memory listShow active GEMINI.md paths/memory list
/directory add <path>Add workspace directory/directory add ./shared-libs
/directory showShow workspace directories/directory show

System & Configuration​

CommandPurposeExample
/settingsOpen settings editor/settings
/themeChange terminal theme/theme
/vimToggle vim-mode editing/vim
/editorSelect default editor/editor
/authChange authentication/auth
/help or /?Show help/help
/aboutShow version info/about

Tools & Integrations​

CommandPurposeExample
/tools [desc]List available tools/tools or /tools read file
/mcpShow MCP servers & tools/mcp
/extensionsList active extensions/extensions

Optimization & Restoration​

CommandPurposeExample
/statsShow token usage/stats
/compressReduce context size/compress
/copyCopy last response/copy
/restore [id]Undo file modifications/restore

Utility​

CommandPurposeExample
/bugReport issue/bug command X doesn't work
/privacyView privacy notice/privacy
/quit or /exitExit Gemini CLI/quit
/initCreate GEMINI.md/init

Deep Dive: Commands You'll Use Most​

1. /chat save & /chat resume — Conversation Checkpointing​

Purpose: Save conversations and resume them later without re-explaining context

Syntax:

/chat save my-project-v1
(Later, in new session...)
/chat resume my-project-v1

When to use:

  • After completing a meaningful chunk of work
  • Before starting something unrelated
  • End of day (save progress)
  • Beginning of day (resume work)

Example Workflow:

You: [Work on authentication for 2 hours]

You: /chat save auth-endpoints-complete
Gemini: ✓ Saved conversation as "auth-endpoints-complete"

(Later, next day...)

You: /chat resume auth-endpoints-complete
Gemini: ✓ Resumed conversation "auth-endpoints-complete"
        [Shows your conversation history]

You: Now let's add password reset functionality
(Continues seamlessly from yesterday)

Why it matters: Without saved conversations, you'd need to re-explain your project every session. With /chat save/resume, context persists.

2. /memory add, /memory show, /memory refresh — Context Management​

Purpose: Manage AI's long-term knowledge about your project

These commands work with GEMINI.md files (covered in Lesson 4).

Syntax:

/memory add we use strict TypeScript and functional patterns
/memory show          # Shows all loaded context
/memory refresh       # Reloads GEMINI.md files
/memory list          # Shows which GEMINI.md files are loaded

How it works:

  1. Gemini CLI looks for GEMINI.md files (global, project, subdirectory)
  2. /memory show displays what's currently loaded
  3. /memory refresh reloads if you update GEMINI.md
  4. /memory add adds temporary facts to the session

Example:

You: /memory show
Gemini CLI:
  Loaded from: ~/.gemini/GEMINI.md
  - My preferences (Python + TypeScript)
  - My coding style

  Loaded from: ./GEMINI.md
  - Project architecture
  - Team standards

You: /memory add API returns ISO 8601 dates
Gemini: ✓ Added to context

3. /stats — Token Budget Awareness​

Purpose: Monitor how many tokens you've used

Syntax:

/stats

Output:

Session Statistics:
- Tokens used: 125,000 / 1,000,000
- Remaining: 875,000
- Session duration: 2h 45m
- Messages: 42

Why it matters: Gemini CLI has a 1 million token context window. Long sessions can approach the limit. /stats tells you when to start a new conversation or use /compress.

4. /compress — Reduce Context Size​

Purpose: Summarize conversation to free up tokens

Syntax:

/compress

What happens:

  • Gemini replaces your conversation with a summary
  • Frees up 40-60% of tokens
  • Keeps key decisions and code context

Example:

You: [After 3 hours of work, 250K tokens used]
You: /stats
Gemini: Used 250,000 / 1,000,000 tokens

You: /compress
Gemini: ✓ Compressed conversation. Now using 90,000 tokens
        (Freed 160,000 tokens for continued work)

5. /clear — Start Fresh​

Purpose: End current conversation and start new one

Syntax:

/clear

Important: Use /chat save before /clear if you want to preserve work.

You: /chat save current-work
Gemini: ✓ Saved

You: /clear
Gemini: ✓ Conversation cleared

You: [Starting completely new task with fresh context]

6. /tools — Discover Available Tools​

Purpose: See what tools Gemini can use

Syntax:

/tools                # List all tools
/tools search         # Filter tools by keyword

Output:

Available Tools:
- GoogleSearch — Search the web
- WriteFile — Create/modify files
- ReadFile — Read file contents
- WebFetch — Fetch URL content
- RunShell — Execute shell commands

7. /mcp — Manage External Integrations​

Purpose: List configured MCP servers and their tools

Syntax:

/mcp

Output:

Configured MCP Servers:
- playwright (✓ connected) — Web browsing
- context7 (✓ connected) — Documentation lookup
- github (✓ connected) — Repository access

8. /directory add & /directory show — Workspace Management​

Purpose: Include additional directories in Gemini's context discovery

Syntax:

/directory add ../shared-libs        # Add a directory
/directory show                      # See all included directories

Why useful: If you work with monorepos or shared libraries, tell Gemini where to find them.

At Commands: Including Files & Directories​

@ Syntax for File References​

Purpose: Include file or directory content in your prompt

Syntax:

@path/to/file.js                # Single file
@./src/                         # Entire directory
@./docs/                        # Read all files in directory
@image.png                      # Image file
@document.pdf                   # PDF file

Examples:

You: @./src/auth.ts explain this authentication flow

You: @./docs/ summarize the project documentation

You: Compare @./old-design.ts and @./new-design.ts

You: Read @./config.json and tell me the API endpoint

Smart Filtering: Gemini CLI automatically excludes:

  • node_modules/
  • .env files
  • .git/ directories
  • Other common non-code files

Shell Commands: System Access​

! Syntax for Shell Execution​

Purpose: Run system commands with AI guidance

Syntax:

!<command>              # Run single command
!                       # Toggle shell mode

Examples:

You: !git status
Gemini: [Executes command, shows output, can discuss results]

You: !
Gemini: [Enters shell mode - all commands execute without !]
$ npm test
$ git log --oneline
$ exit                  # Exit shell mode back to chat

Safety: Gemini can warn you about dangerous commands before executing.

Keyboard Shortcuts​

Quick actions while typing:

ShortcutAction
Ctrl+LClear terminal (same as /clear)
Ctrl+ZUndo last input
Ctrl+Shift+ZRedo undone input

Practical Daily Workflow​

Here's how these commands flow in real work:

Morning: Resume Previous Work​

$ gemini
Gemini: Loading context from ~/.gemini/GEMINI.md and ./GEMINI.md

You: /chat resume api-endpoints-in-progress
Gemini: ✓ Resuming conversation
        (Shows your conversation history from yesterday)

You: Let's continue with the user endpoints
(Continues exactly where you left off)

Mid-Day: Monitor Budget​

You: [After 2 hours of work]

You: /stats
Gemini: Used 200,000 / 1,000,000 tokens remaining

You: We're in good shape, let's keep going

Switch Projects​

You: /chat save api-endpoints-complete
Gemini: ✓ Saved

You: /clear
Gemini: ✓ Fresh conversation started

You: /chat resume frontend-redesign
Gemini: ✓ Resuming previous project

End of Day: Save Progress​

You: /chat save current-work-session
Gemini: ✓ Saved

You: Good progress today. See you tomorrow!
(Exit Gemini CLI)

Next Day: Resume Seamlessly​

$ gemini

You: /chat resume current-work-session
Gemini: ✓ Loaded. Where were we...

Common Questions​

Q: What's the difference between /chat save and GEMINI.md?

A:

  • /chat save = Saves your CONVERSATION (what you discussed)
  • GEMINI.md = Persistent PROJECT CONTEXT (what Gemini should know about your project)

Use both together:

  • GEMINI.md tells Gemini about your architecture
  • /chat save remembers what you've discussed this session

Q: Can I use /chat resume to work with someone else's saved conversation?

A: Not directly. Saved conversations are personal. But you can use /chat share file.md to export, then share the file.

Q: If I use /compress, will I lose my conversation?

A: No. It summarizes your conversation but keeps important decisions and context. You can still reference previous work.

Q: What happens if I close Gemini without using /chat save?

A: Your conversation is lost. Always save important work with /chat save.

Q: Can I use shell commands safely?

A: Yes. Gemini will warn you about dangerous commands (rm, sudo, etc.) before executing. Always read the warning before confirming.

Exercises​

Exercise 1: Save Your First Conversation​

Have a real conversation about something meaningful:

$ gemini
You: [Work on a real task for 20+ minutes]
You: /chat save my-first-saved-conversation
Gemini: ✓ Saved

Expected: You can now list and resume this conversation anytime.

Exercise 2: List and Inspect​

You: /chat list
(See all saved conversations)

You: /chat save another-one
You: /chat list
(See the new conversation)

Exercise 3: Resume and Continue​

Close Gemini CLI completely. Then:

$ gemini
You: /chat resume my-first-saved-conversation

Expected: Your conversation history is restored.

Exercise 4: Check Your Token Budget​

You: /stats
(See how many tokens you've used)

You: (Work for 10 more minutes)

You: /stats
(See tokens increased)

Exercise 5: Reference a File​

Create a simple file:

$ echo "console.log('hello')" > test.js

Then in Gemini:

You: @test.js what does this code do?
Gemini: (Shows file content and explains)

Key Takeaways​

  • 29+ slash commands provide meta-level control (don't memorize—reference as needed)
  • /chat save and /chat resume are your superpowers for conversation persistence
  • /memory show and /memory refresh work with GEMINI.md for project context (covered next)
  • /stats and /compress keep you aware of token budget
  • @ syntax includes file content without typing /copy-paste
  • ! syntax runs shell commands with AI guidance
  • Commands are discoverable—/help shows everything

Next lesson: You'll learn GEMINI.md context files, so Gemini automatically understands your project architecture and coding standards.