Skip to content

Mode Switching (Chat / Execute)

YAICLI offers multiple interaction modes to suit different needs. This page explains each mode and how to switch between them.

Available Modes

YAICLI has three primary interaction modes:

1. Chat Mode (💬)

Chat mode provides a natural conversation experience with the LLM, maintaining context and formatting responses with markdown.

Features: - Full conversation history and context tracking - Markdown and code formatting - Reasoning display for complex queries - Persistent chat storage and management

Best for: - Extended conversations - Complex questions requiring context - Learning and exploration

2. Execute Mode (🚀)

Execute mode specializes in generating and safely running shell commands based on natural language descriptions.

Features: - OS-specific command generation - Command review and editing before execution - Command output display - Safe execution with confirmation

Best for: - System administration tasks - File operations - Command discovery - Automating repetitive tasks

3. Quick Query Mode (Default)

The default non-interactive mode gives you quick answers without entering a persistent session.

Features: - Direct answers to questions - Support for piped input - Fast responses with minimal overhead

Best for: - One-off questions - Quick information lookups - Script integration

Switching Between Modes

Command Line Switches

Use these flags when starting YAICLI:

# Start in chat mode
ai --chat "Optional chat title"

# Start in execute mode
ai --shell "Create a backup script"

# Quick query (default mode)
ai "What is the capital of France?"

Interactive Mode Switching

When in interactive mode (chat or execute), you can switch modes in several ways:

Using Tab Key

Press the Tab key to toggle between Chat and Execute modes.

Using Commands

Type /mode followed by the mode name:

/mode chat   # Switch to chat mode
/mode exec   # Switch to execute mode

Visual Indicators

The prompt changes to indicate your current mode: - Chat mode: 💬 > - Execute mode: 🚀 >

Mode Behavior

Chat Mode Behavior

In chat mode: - The AI responds conversationally with rich formatting - Context is maintained between interactions - Input is treated as a conversation prompt

Example: 

💬 > Tell me about quantum computing
Assistant:
Quantum computing uses quantum bits (qubits) that can exist in multiple 
states simultaneously due to superposition...

Execute Mode Behavior

In execute mode: - The AI interprets input as a request for a command - Generates appropriate commands for your OS and shell - Displays the command for review before execution

Example: 

🚀 > Find all PDF files in my Downloads folder
Assistant:
find ~/Downloads -name "*.pdf"
╭─ Suggest Command ─────────────────────╮
│ find ~/Downloads -name "*.pdf"        │
╰─────────────────────────────────────────╯
Execute command? [e]dit, [y]es, [n]o (n):

Chat Persistence

When starting a chat session with a title, the conversation is automatically saved:

ai --chat "Python tips"

A chat started without a title is treated as temporary but can be saved later:

💬 > /save "Python tips"
Chat saved as: Python tips

You can list, load, and manage saved chats:

💬 > /list
Saved chats:
1. Python tips (2023-07-01)
2. Work project (2023-06-28)

💬 > /load 1
Loading chat: Python tips

Next Steps