Skip to main content

Documentation Index

Fetch the complete documentation index at: https://docs.argentos.ai/llms.txt

Use this file to discover all available pages before exploring further.

​
Overview

Slash commands are shorthand actions you can invoke from any channel by prefixing a message with /. They provide quick access to common agent functions without needing to phrase a natural language request.

​
Available Commands

CommandDescription
/statusShow agent status and system health
/tasksList current tasks
/task <title>Create a new task
/memory <query>Search agent memory
/forget <id>Delete a memory entry
/modelShow current model and tier
/model <name>Switch to a specific model
/clearClear the current session context
/helpShow available commands

​
Using Slash Commands

Send a slash command from any connected channel: 
/tasks
The agent recognizes the slash prefix and executes the command directly instead of treating it as a conversational message.

​
With Arguments

Some commands accept arguments: 
/task Review the Q4 security audit report
/memory Jason's infrastructure specs
/model claude-sonnet-4-20250514

​
Dashboard Integration

Slash commands also work in the dashboard chat input. The dashboard may render command results with special formatting (tables, status indicators) that channels display as plain text.

​
Custom Slash Commands

Plugins and skills can register custom slash commands: 
{
  "commands": [
    {
      "name": "deploy",
      "description": "Deploy the latest build",
      "handler": "handlers/deploy.js"
    }
  ]
}

​
Command vs Conversation

The agent distinguishes between slash commands and natural language:
InputBehavior
/tasksExecutes the tasks command directly
Show me my tasksThe agent interprets this conversationally and may use the tasks tool
What's on my task list for today?Natural language, agent decides how to respond
Both approaches reach the same underlying tools, but slash commands skip the LLM reasoning step for faster execution.

​
Permissions

Slash commands respect the same tool policies as regular tool use. If exec is restricted, the /exec command (if it existed) would also be restricted.