Terminal Tab

The Terminal tab gives you a real shell inside the Agent's sandbox environment. You can run commands, install packages, inspect files, and debug scripts — all without leaving the Buda interface.

Opening a terminal

Click the + button in the tab bar and select Terminal. A new terminal tab opens and connects to the Agent's sandbox within a few seconds.

If the sandbox is not yet running, Buda starts it automatically. The first connection may take 10–20 seconds while the environment initializes.

Running commands

Type any shell command and press Enter. The terminal supports the full POSIX shell environment:

# Navigate the file system
ls -la /workspace

# Run a script
python3 scripts/process_data.py

# Install a package
pip install pandas

# Check running processes
ps aux

The working directory starts at the Agent's Drive root (/workspace by default).

Splitting the terminal

You can open multiple terminal panes side by side. Click the + button in the tab bar, select Terminal, and a second pane appears next to the first. Each pane is an independent shell session.

To close a pane, click the × button in the top-right corner of that pane.

Persistent sessions

The terminal session persists as long as the sandbox is running. If you switch to another tab and come back, the shell history and running processes are still there.

If the sandbox goes to sleep (after the idle timeout), the terminal reconnects automatically when you return.

Environment variables

Environment variables set in the Agent's settings are available in the terminal automatically. You can also set them inline:

export API_KEY=your_key_here
python3 run.py

Variables set this way are scoped to the current shell session.

File system access

The terminal has full read/write access to the Agent's Drive. Files you create or modify in the terminal appear immediately in the Drive tab.

# Create a file
echo "hello world" > /workspace/notes.txt

# Edit with nano or vim
nano /workspace/config.yaml

Watching the Agent use the terminal

When the Agent executes terminal commands as part of a task, you can see the exact commands and output in the chat panel. Switch to the Terminal tab to see the live shell state after the Agent finishes.

Tips

• Use Ctrl+C to interrupt a running command
• Use Ctrl+L to clear the screen
• Use Tab for auto-completion
• The terminal supports ANSI colors and most interactive CLI tools

Related pages

• Agent Workspace Overview
• Drive Tab
• Git Tab