Understanding the Interface
Learn how to navigate the Claude Code terminal interface effectively.
The Main Prompt
When you start Claude Code, you'll see the main interactive prompt:
claude > _
Type your requests here in natural language. Claude understands context from previous messages in the conversation.
Tool Usage Indicators
When Claude uses tools, you'll see indicators for:
- File reads — Shows which files Claude is examining
- File writes — Shows proposed changes with diffs
- Command execution — Shows commands being run and their output
- Web searches — Shows search queries and results
Permission System
Claude Code uses a tiered permission system:
Automatic Actions
- Reading files
- Searching code
- Analyzing project structure
Actions Requiring Approval
- Writing or editing files
- Running shell commands
- Installing packages
Permission Modes
| Mode | Description |
|---|---|
| Default | Ask for approval on writes and commands |
| Plan | Require plan approval before any changes |
| Auto-accept | Automatically approve all actions |
Keyboard Shortcuts
| Shortcut | Action |
|---|---|
Ctrl+C | Cancel current operation or exit |
Ctrl+L | Clear the screen |
Up/Down | Navigate command history |
Tab | Autocomplete file paths |
Status Bar
The status bar is a single line pinned below your input prompt. When configured with the ccstatusline package (see System Configuration), it looks like this:
Model: Opus 4.5 | Ctx: 31.7k | ⌐main | (+4,-0)
- Model — which model is powering the session (Opus 4.5, Sonnet, etc.)
- Ctx — tokens of context used so far, the most important number to watch
- Branch — current git branch
- Diff — uncommitted lines added/removed in your working tree
The context counter is your fuel gauge. When it gets high (~60%+), consider running /handoff to capture progress and start a fresh session before quality starts to drop.
Next Steps
Now that you understand the interface, learn about CLAUDE.md Configuration to customize Claude Code for your projects.