Skip to Content
Commands vs Slash Commands

Commands vs Slash Commands

In AgentLoop’s interactive mode, there are two distinct ways to interact with the system: natural language commands and slash commands. Understanding when to use each is key to an efficient workflow.

Overview

TypeFormatPurposeExample
Natural LanguageJust type and press EnterConversational interaction with agentsBreak down user auth into tasks
Slash CommandsStart with /System operations and navigation/orchestrator run

Natural Language Commands

Natural language commands are the primary way to interact with AgentLoop. You simply type what you want in plain English (or any language), and AgentLoop’s AI agents interpret and execute your request.

How It Works

When you type a message without a / prefix:

  1. Your message is sent to the orchestrator agent
  2. The agent interprets your intent
  3. Tool calls are made behind the scenes to:
    • Update the kanban board
    • Manage tasks and dependencies in the DAG
    • Delegate work to specialized agents
  4. Results are streamed back to you

What You Can Do

Task Management

> Add a new task for implementing the login page > Mark task 5 as in-progress > Move all completed tasks to done > What's blocking task 7?

Agent Delegation

> Have the engineer agent implement the API endpoint > Ask the QA agent to test the authentication flow > Let the code analyzer look at the payment module

DAG and Dependencies

> Add a dependency so task 3 must complete before task 7 > Show me which tasks are ready to start > Remove the dependency between tasks 2 and 5

Board Operations

> Show me the kanban board > What tasks are currently in review? > Summarize the project status

Planning and Breakdown

> Break down "user authentication" into tasks and add them to the board > Create a sprint plan for the checkout feature > Generate user stories for the admin dashboard

Advantages

  • Intuitive - No syntax to memorize
  • Flexible - Describe what you want in your own words
  • Contextual - Agents understand the conversation history
  • Powerful - Complex multi-step operations in a single request

Slash Commands

Slash commands are prefixed with / and provide direct access to specific system operations. They’re deterministic—each command does exactly one thing.

How It Works

When you type a message starting with /:

  1. The command is parsed directly (no AI interpretation)
  2. The corresponding system function executes
  3. Results or views are displayed

Categories of Slash Commands

System Commands

/help # Show all available commands /exit # Exit interactive mode /clear # Clear conversation history /init # Initialize AgentLoop for this project

Orchestrator Control

/orchestrator run # Start processing tasks /orchestrator run --infinite # Run continuously /orchestrator stop # Stop the orchestrator /orchestrator status # Check orchestrator status

Views and Visualization

/orchestrator kanban # Open kanban board /orchestrator agents # Live agent monitoring /orchestrator dag ascii # ASCII DAG visualization

Task Queries

/tasks # List all tasks /tasks --status in-progress # Filter by status /tasks --priority high # Filter by priority

Configuration

/config show # Show current config /settings show # Show settings /settings telemetry off # Disable telemetry

Advantages

  • Predictable - Same command, same result every time
  • Fast - No AI processing overhead
  • Specific - Access exact functionality
  • Scriptable - Easy to document and repeat

When to Use Each

Use natural language when you want to:

  • Create or modify tasks - “Add a task for the login page”
  • Delegate to agents - “Have the engineer implement this”
  • Manage dependencies - “Task 3 should wait for task 2”
  • Ask questions - “What’s the status of the API work?”
  • Plan features - “Break this feature into tasks”
  • Complex operations - “Move all high-priority blocked tasks to todo and assign them to the engineer”

Natural language is best when your request involves:

  • Multiple steps
  • Agent coordination
  • Contextual understanding
  • Creative or open-ended work

Comparison Examples

Here’s the same goal achieved both ways:

Viewing Tasks

> Show me all the tasks that are currently in progress

The agent interprets your request, queries the task system, and provides a formatted response with context.

Starting the Orchestrator

> Start processing the tasks, run continuously until I stop you

The agent understands you want continuous execution and starts the orchestrator in infinite mode.

Adding Tasks

> I need to add user authentication with OAuth support, password reset, > and remember me functionality. Break these down into tasks.

The PM agent analyzes requirements, creates multiple well-structured tasks with descriptions, and organizes them on the board.

Mixing Both Approaches

In practice, you’ll use both in the same session. Here’s a typical workflow:

# Initialize with slash command > /init # Describe what you want naturally > I need to build a REST API for user management with CRUD operations # View the generated tasks > /orchestrator kanban # Ask follow-up questions naturally > Can you add authentication middleware to this? # Start processing with slash command > /orchestrator run # Check status naturally > How's it going? Any tasks blocked? # Stop when needed > /orchestrator stop

Quick Reference

I want to…Use
Add/modify tasksNatural language
Delegate to agentsNatural language
Manage dependenciesNatural language
Ask questionsNatural language
Start/stop orchestratorSlash command
Open kanban boardSlash command
Exit the appSlash command
Initialize projectSlash command
Change settingsSlash command

Pro tip: When in doubt, try natural language first. If you need more control or a specific system function, use the corresponding slash command.

Last updated on