Skip to main content

​
Overview

Qwen Code’s interactive mode provides a rich terminal UI for conversing with AI in real-time. When you run qwen without arguments, you enter an interactive session where you can chat naturally, issue commands, and see AI actions unfold live.

​
Starting Interactive Mode

# Launch interactive UI
qwen

# Start with a prompt and continue interactively
qwen --prompt "Explain this codebase" --continue

​
Key Features

​
Real-Time Streaming

The interactive UI streams responses as they’re generated, showing:
  • AI thinking and reasoning
  • Tool calls being executed
  • File edits with syntax highlighting
  • Command outputs with proper formatting

​
Visual Feedback

When AI executes tools, you see:
  • Tool name and arguments
  • Progress indicators for long-running operations
  • Success/failure status with colored indicators
  • Structured output formatting

​
Terminal Controls

​
Keyboard Shortcuts

ShortcutAction
Ctrl+CCancel current operation
Ctrl+DExit session
Shift+TabCycle through approval modes
EscCancel pending slash command
↑ / ↓Navigate command history
TabAuto-complete commands and paths

​
Vim Mode

Enable Vim-style keybindings for input: 
/vim
In Vim mode:
  • Esc enters normal mode
  • i, a, o enter insert mode
  • h, j, k, l for navigation
  • w, b, e for word movement

​
Session Management

​
Starting Fresh

# Clear conversation and start new session
/clear

# Alternative commands
/reset
/new
/clear resets the conversation history and frees up context, but preserves your configuration and settings.

​
Resuming Sessions

# Open session picker
/resume
The resume dialog shows:
  • List of recent sessions with timestamps
  • Session descriptions (first user message)
  • Fuzzy search to filter sessions
  • Arrow keys to select, Enter to load

​
Exiting Gracefully

# Quit with summary
/quit

# Alternative
/exit
On exit, you’ll see:
  • Total session duration
  • Number of turns
  • Token usage statistics

​
Advanced UI Features

​
Debug Mode

Enable detailed logging: 
export DEBUG=qwen:*
qwen
Debug mode displays:
  • API request/response details
  • Tool execution traces
  • Performance metrics
  • Internal state changes

​
Stats and Monitoring

# View session statistics
/stats

# Model-specific stats
/stats model

# Tool usage breakdown
/stats tools

​
Theme Customization

# Open theme selector
/theme
Available themes:
  • Light
  • Dark
  • High Contrast
  • Solarized
  • Custom themes via settings

​
Rich Formatting

The interactive UI renders:
  • Headers and emphasis
  • Code blocks with syntax highlighting
  • Lists (ordered and unordered)
  • Tables
  • Blockquotes
  • Links (not clickable, but visible)

​
Context Indicators

The footer displays:
  • Current approval mode (plan/default/auto-edit/yolo)
  • Token usage (current/limit)
  • Active model name
  • Session ID
  • Working directory
Press Shift+Tab to cycle approval modes on-the-fly without opening the settings dialog.

​
Accessibility

​
Screen Reader Support

Enable screen reader mode: 
// settings.json
{
  "accessibility": {
    "screenReader": true,
    "enableLoadingPhrases": true
  }
}
Screen reader mode:
  • Announces AI responses
  • Reads tool execution status
  • Provides audio feedback for state changes
  • Simplifies visual output for clarity

​
Keyboard Navigation

All features are keyboard-accessible:
  • Tab through interactive elements
  • Arrow keys for list navigation
  • Enter to confirm, Esc to cancel
  • No mouse required

​
Performance Considerations

​
Large Outputs

The UI handles large outputs efficiently:
  • Virtual scrolling for long conversations
  • Pagination for tool results
  • Compression of old context
  • Memory-efficient rendering

​
Memory Management

# Compress context to free memory
/compress

# Alternative
/summarize
Compression:
  • Replaces old messages with summaries
  • Preserves semantic meaning
  • Reduces token usage
  • Allows longer sessions

​
Troubleshooting

Try:
  • Compress context with /compress
  • Start a new session with /clear
  • Reduce model token limits in settings
  • Disable debug mode if enabled
  • Ensure terminal supports UTF-8
  • Resize terminal window
  • Try a different theme with /theme
  • Check for font issues
  • Verify terminal supports 256 colors
  • Try export TERM=xterm-256color
  • Use high contrast theme
  • Check terminal emulator settings
  • Check for terminal keybinding conflicts
  • Disable Vim mode if enabled
  • Try alternative shortcuts
  • Verify terminal emulator compatibility

​
Best Practices

  1. Use /clear periodically to free up context and maintain performance
  2. Enable Vim mode if you prefer modal editing
  3. Customize your theme for optimal readability
  4. Use /stats to monitor token usage and stay within limits
  5. Leverage /compress before hitting context limits
  6. Resume sessions to continue previous work
  7. Use shortcuts to speed up common operations

​
Next Steps

Session Commands

Learn all available slash commands

Approval Modes

Control how AI tools are approved

Headless Mode

Run non-interactively for automation

Subagents

Delegate tasks to specialized agents