Skip to main content

​
Core Commands

​
Conversation Management

​
/new (alias: n)

Start a new conversation while preserving history. 
/new
This creates a fresh conversation context while maintaining access to previous conversations through /conversation.

​
/conversation or /conversations (alias: c)

List and switch between all conversations for the active workspace. 
/conversation
Displays an interactive list of conversations with timestamps and allows you to switch between them.

​
/compact

Compact the conversation context to reduce token usage. 
/compact
Use this when approaching context limits. Forge will compress the conversation history while preserving important information.

​
Agent Switching

​
/forge or /act

Switch to Forge agent (implementation mode) with code changes enabled. 
/forge
> Implement a new login feature

​
/muse or /plan

Switch to Muse agent (planning mode) without code changes. 
/muse
> Help me design the architecture for this feature

​
/sage

Switch to Sage agent for systematic codebase exploration and analysis. 
/sage
> Analyze how authentication is implemented

​
/help

Switch to Help agent for questions about Forge itself. 
/help
> How do I configure MCP servers?

​
/agent (alias: a)

Interactively select and switch between available agents. 
/agent
Displays a menu of all configured agents including custom agents from your workflow.

​
Configuration

​
/model (alias: m)

Switch to a different AI model. 
/model
Displays an interactive list of available models for your current provider, showing context length and capabilities.

​
/provider (alias: p)

Switch to a different AI provider. 
/provider
Shows all configured providers (OpenAI, Anthropic, etc.) and allows you to switch between them.

​
/login

Configure provider credentials interactively. 
/login
Guides you through adding or updating provider API keys and credentials.

​
/logout

Remove provider credentials. 
/logout
Removes stored credentials for the selected provider.

​
Information Commands

​
/info (alias: i)

Display system and session information. 
/info
Shows current agent, model, provider, workspace, and other session details.

​
/env (alias: e)

Display environment configuration. 
/env
Shows environment variables and configuration settings affecting Forge.

​
/usage

Display usage statistics including tokens and requests. 
/usage
Shows token usage, API requests, and cost information for the current session.

​
/tools (alias: t)

List all available tools with descriptions and schema. 
/tools
Displays all tools available to the AI agent, including built-in tools and MCP tools.

​
Git Operations

​
/commit

Generate an AI commit message and commit changes. 
/commit
Analyzes staged changes and generates a meaningful commit message, then commits. With max diff size: 
/commit 5000
Limits the diff size analyzed to 5000 bytes for very large changesets.
Forge analyzes your git diff to generate contextual commit messages. Make sure you have changes staged with git add before using this command.

​
/retry (alias: r)

Retry the last command without modifying context. 
/retry
Useful when a command fails due to network issues or temporary errors.

​
Export and Debugging

​
/dump (alias: d)

Save the current conversation to a file. JSON format (default): 
/dump
Saves conversation as JSON file in the current directory. HTML format: 
/dump html
Saves conversation as a formatted HTML file for easy viewing in browsers.

​
Workspace Management

​
/index

Index the current workspace for semantic code search. 
/index
Creates a semantic index of your codebase, enabling more intelligent code search and context retrieval.

​
Application Control

​
/update

Update Forge to the latest compatible version. 
/update
Checks for and installs the latest version of Forge.

​
/exit

Exit Forge and return to your shell. 
/exit
You can also use Ctrl+D to exit.

​
Shell Commands

Commands starting with ! are executed directly in your shell: 
!git status
!npm test
!cargo build
!ls -la
Shell commands run in your current working directory and have access to your full shell environment.

​
Custom Commands

Custom commands can be defined in your forge.yaml: 
commands:
  - name: "refactor"
    description: "Refactor selected code"
    prompt: "Please refactor this code to improve readability"
Use them like built-in commands: 
/refactor

​
Agent Commands

When you configure custom agents, Forge generates commands automatically: 
/agent-custom-name     # 🤖 Switch to Custom Name agent
/agent-data-analyst    # 🤖 Switch to Data Analyst agent

​
Workflow Commands

Workflow commands appear with the ⚙ prefix: 
/deploy        # ⚙ Deploy the application
/run-tests     # ⚙ Run test suite

​
Command Aliases

Many commands have shorter aliases for faster typing:
CommandAlias
/newn
/infoi
/enve
/modelm
/providerp
/conversationc
/toolst
/dumpd
/retryr
/agenta

​
Command Completion

Press Tab after typing / to see all available commands with descriptions. The completion menu shows:
  • Built-in system commands
  • Custom commands from forge.yaml
  • Agent switching commands
  • Workflow commands

​
Tips

  • Use /info to check your current configuration before making changes
  • Use /compact when you see warnings about context length
  • Combine shell commands with natural language: run !npm test then ask “what failed?”
  • Use /dump html to save important conversations for documentation
  • Set up custom commands for frequently used prompts

​
Next Steps

Build docs developers (and LLMs) love

Get started for freeTalk to us