AI Guide

0

0

History

AI Assistant Guide

SkynetBot features a powerful AI assistant with multi-provider LLM support, conversation memory, and advanced configuration options.

Overview

The AI system provides:

  • Multi-Provider Support: OpenAI (GPT-4, GPT-4o), Anthropic (Claude), Groq (Llama, Mixtral), Ollama (local models)
  • Conversation Memory: Remembers context within conversations
  • Web Search: Integrated web search for up-to-date information
  • Rate Limiting: Configurable cooldowns to prevent abuse
  • Usage Tracking: Monitor token usage and costs
  • Budget Controls: Set daily limits per user or server
  • Custom Personalities: Configure custom system prompts

Commands

Command Description
!ai ask <message> Chat with the AI assistant
!ai stream <message> Chat with streaming response
!ai clear Clear your conversation memory
!ai search <query> Search the web using AI
!ai variables Show available template variables
!ai stats View usage statistics (admin)

Basic Usage

Simply ask the AI anything:

!ai ask What is the capital of France?
!ai ask Explain quantum computing in simple terms
!ai ask Write a haiku about Discord

Streaming Responses

For longer responses, use streaming to see the response as it's generated:

!ai stream Tell me a story about a brave knight

Search the web for current information:

!ai search Latest news about artificial intelligence
!ai search Weather in New York today

Clearing Memory

Clear your conversation history to start fresh:

!ai clear

Template Variables

Use these variables in your messages for dynamic content:

Variable Description
{{user}} Your username
{{user.id}} Your Discord user ID
{{user.mention}} Mentions you in the response
{{channel}} Current channel name
{{server}} Server name
{{date}} Current date
{{time}} Current time
{{datetime}} Current date and time

Example:

!ai ask Hello {{user}}, what time is it in {{channel}}?

Supported Providers

OpenAI

  • Models: GPT-4o, GPT-4o-mini, GPT-4, GPT-3.5-turbo
  • Best for: General-purpose tasks, coding, creative writing
  • Requires: API key

Anthropic (Claude)

  • Models: Claude 3.5 Sonnet, Claude 3 Opus, Claude 3 Sonnet, Claude 3 Haiku
  • Best for: Long-form content, analysis, nuanced conversations
  • Requires: API key

Groq

  • Models: Llama 3.1, Mixtral
  • Best for: Fast responses, open-source models
  • Requires: API key

Ollama

  • Models: Any locally hosted model
  • Best for: Privacy-focused, offline use, custom models
  • Requires: Local Ollama installation

Server Configuration (Admins)

Server administrators can configure the AI system through the web dashboard:

Dashboard → AI Settings

  • Provider & Model: Choose default provider and model
  • System Prompt: Customize the AI's personality
  • Rate Limits: Set cooldowns and usage limits

Dashboard → AI Governance

  • Model Policy: Allow or deny specific models
  • Tool Access: Control which AI tools are available
  • Budget Limits: Set daily token/cost limits

Dashboard → AI Memory

  • History Limit: How many messages to remember
  • Per-User Memory: Enable individual memory per user

Dashboard → AI Personality

  • System Prompt: Define the AI's character and behavior
  • Response Style: Configure how the AI responds

Rate Limits

To prevent abuse, the AI system has built-in rate limiting:

  • Cooldown: Minimum time between requests (default: 5 seconds)
  • Per-User Limit: Maximum requests per user per day
  • Per-Channel Limit: Maximum requests per channel per hour
  • Token Budget: Maximum tokens per user/server per day

Administrators can adjust these limits in the dashboard.

Best Practices

For Users

  1. Be specific: Clear questions get better answers
  2. Provide context: Include relevant background information
  3. Use streaming: For long responses, use !ai stream
  4. Clear when needed: Use !ai clear if the AI seems confused

For Administrators

  1. Set reasonable limits: Balance accessibility with abuse prevention
  2. Monitor usage: Check !ai stats regularly
  3. Customize personality: Tailor the AI to your server's needs
  4. Use governance: Restrict expensive models if needed

Troubleshooting

"Rate limited"

You're sending requests too quickly. Wait for the cooldown period.

"Budget exceeded"

You've reached your daily token limit. Wait until tomorrow or ask an admin to increase limits.

"Provider error"

The AI provider is experiencing issues. Try again later or contact an admin.

"Model not available"

The requested model is not configured or has been restricted by administrators.

See also: Commands | Server Owner Guide | FAQ