Skip to main content

​
What is Interactive Mode?

Forge’s interactive mode is a powerful terminal interface that allows you to interact with AI agents through natural language and commands. When you run forge without arguments, you enter interactive mode where you can:
  • Chat with AI agents using natural language
  • Execute slash commands for system operations
  • Run shell commands directly with the ! prefix
  • Switch between different AI agents and models
  • Manage conversations and view usage statistics

​
Types of Input

Forge supports three types of input in interactive mode:

​
1. Natural Language Messages

Any text that doesn’t start with / or ! is treated as a message to the AI agent: 
> Can you help me refactor this authentication module?

​
2. Slash Commands

Commands that start with / are system commands that control Forge’s behavior: 
/new        # Start a new conversation
/model      # Switch AI models
/help       # Get help information
See the Slash Commands reference for a complete list.

​
3. Shell Commands

Commands that start with ! are executed directly in your shell: 
!git status
!npm test
!ls -la

​
Agent Modes

Forge includes several built-in agents, each optimized for different tasks:

​
Forge Agent (/forge or /act)

Implementation mode - Makes code changes and executes operations: 
/forge
> Add error handling to the login function

​
Muse Agent (/muse or /plan)

Planning mode - Analyzes and plans without making changes: 
/muse
> How should I architect this new feature?

​
Sage Agent (/sage)

Research mode - Systematic codebase exploration and analysis: 
/sage
> Analyze the authentication flow in this codebase

​
Help Agent (/help)

Tool assistance - Answers questions about Forge itself: 
/help
> How do I configure a custom provider?

​
Interactive Features

​
Tab Completion

Press Tab to autocomplete commands, file paths, and agent names. If multiple options are available, a completion menu will appear.

​
Command History

Forge maintains a history of your commands:
  • Press Ctrl+R to search through command history
  • Use arrow keys to navigate through previous commands
  • History is stored in your system config directory

​
Multi-line Input

Press Alt+Enter to insert a newline without submitting: 
> This is a multi-line
  prompt that spans
  several lines

​
Context Management

Forge automatically manages conversation context:
  • Use /compact to compress context when approaching limits
  • Use /new to start fresh while preserving history
  • Use /conversation to switch between conversations

​
Custom Commands

You can define custom commands in your forge.yaml file: 
commands:
  - name: "refactor"
    description: "Refactor selected code"
    prompt: "Please refactor this code to improve readability and performance"
  - name: "review"
    description: "Code review"
    prompt: "Review this code for bugs and improvements"
Then use them with: 
/refactor
/review

​
Agent Commands

When you have custom agents configured, Forge automatically generates commands for switching to them: 
/agent-custom-name    # Switch to agent named "custom-name"
/agent-data-analyst   # Switch to agent named "data-analyst"
Use /agent to interactively select from available agents.

​
Workflow Commands

Workflow commands defined in your workflow configuration appear with a ⚙ prefix: 
/deploy       # ⚙ Deploy the application
/run-tests    # ⚙ Run the test suite

​
Next Steps

Build docs developers (and LLMs) love

Get started for freeTalk to us