Configuration & Usage

Complete guide to configuring and using Daia.

Table of Contents

  1. Environment Variables
    1. Required Variables
    2. Optional Variables
    3. Configuration Details
  2. System Prompt Customization
    1. Example System Prompts
    2. Best Practices
  3. Usage Guide
    1. Starting a Conversation
    2. Continuing Conversations
    3. Starting Fresh
  4. Features in Detail
    1. Markdown Table Rendering
    2. Smart Message Chunking
    3. Google Search Integration
    4. User Recognition
  5. Advanced Configuration
    1. Multiple Servers
    2. Channel-Specific Behavior
    3. Rate Limiting
    4. Performance Tuning
  6. Common Use Cases
    1. Community Support Bot
    2. Personal Assistant
    3. Multilingual Server
  7. Troubleshooting
    1. Bot responds to every message
    2. Bot doesn’t remember context
    3. Responses are too random/inconsistent
    4. Tables not rendering
  8. Next Steps

Environment Variables

All configuration is done through environment variables in the .env file.

Required Variables

Variable Description Example
DISCORD_BOT_TOKEN Your Discord bot token from the Developer Portal YOUR_DISCORD_BOT_TOKEN_HERE
CHAT_MODEL_API_KEY Your Google Gemini API key AIzaSyD1234567890abcdefghijklmnopqrstuv
CHAT_MODEL The Gemini model to use gemini-1.5-flash or gemini-1.5-pro
CHAT_TEMPERATURE Controls response randomness (0.0–2.0) 1.0
HISTORY_LIMIT Maximum number of messages to fetch from history 50
CHAT_SYS_PROMPT_PATH Path to the system prompt file config/chat_sys_prompt.txt

Optional Variables

Variable Description Default Example
DISCORD_BOT_ACTIVITY Activity status displayed for the bot None Surfing, Listening to music
ALLOWED_CHANNELS Comma-separated list of channel IDs where bot responds without mentions None 123456789,987654321
ENABLE_CONTEXTUAL_SYSTEM_PROMPT Enable user recognition by display name off on (recommended)
CHAT_MODEL_PROVIDER LLM provider to use gemini gemini

Configuration Details

CHAT_MODEL

Choose based on your needs:

  • gemini-1.5-flash: Faster responses, lower cost, good for most use cases
  • gemini-1.5-pro: More capable, better for complex tasks, higher cost

CHAT_TEMPERATURE

Controls the randomness of responses:

  • 0.0: Deterministic, focused responses
  • 1.0: Balanced (recommended)
  • 2.0: More creative and varied responses

HISTORY_LIMIT

Number of previous messages to include as context:

  • Lower values (10-20): Faster, less context
  • Medium values (30-50): Balanced (recommended)
  • Higher values (50-100): More context, slower, higher token usage

ALLOWED_CHANNELS

If set, the bot will respond to all messages in these channels without requiring a mention. To find a channel ID:

  1. Enable Developer Mode in Discord (User Settings → Advanced → Developer Mode)
  2. Right-click a channel and select “Copy Channel ID”

Example:

ALLOWED_CHANNELS=1234567890,9876543210,1122334455

ENABLE_CONTEXTUAL_SYSTEM_PROMPT

When enabled (on), the bot will:

  • Address users by their Discord display name
  • Provide more personalized responses
  • Include user context in the system prompt

Recommended setting: on

System Prompt Customization

The system prompt defines your bot’s personality and behavior. Edit config/chat_sys_prompt.txt to customize it.

Example System Prompts

Helpful Assistant:

You are Daia, a helpful AI assistant in a Discord server.
You provide clear, accurate information and maintain a friendly, professional tone.
When you don't know something, you admit it rather than making up information.

Casual Friend:

You are Daia, a chill AI friend hanging out in Discord.
You keep things casual and fun, using emojis occasionally 😊
You're knowledgeable but never condescending.

Technical Expert:

You are Daia, a technical AI assistant specializing in programming and technology.
You provide detailed, accurate technical information with code examples when relevant.
You explain complex concepts clearly and concisely.

Best Practices

  • Keep it concise (2-5 sentences)
  • Define the bot’s role and personality
  • Set expectations for response style
  • Include any specific guidelines or restrictions

Usage Guide

Starting a Conversation

There are three ways to interact with Daia:

1. In Allowed Channels

If you’ve configured ALLOWED_CHANNELS, simply type your message:

What's the weather like today?

2. Mention the Bot

In any channel where the bot has access:

@Daia what's the capital of France?

3. Direct Message

Send the bot a DM for private conversations:

Can you help me with something?

Continuing Conversations

Daia remembers recent message history (up to HISTORY_LIMIT). You can continue the conversation naturally:

You: What's Python?
Daia: Python is a high-level programming language...

You: What's it used for?
Daia: Python is commonly used for...

Starting Fresh

Use the /newchat slash command to clear the conversation history and start fresh:

/newchat

This is useful when:

  • Switching topics
  • The bot seems confused by previous context
  • You want to reset the conversation

Features in Detail

Markdown Table Rendering

When Daia’s response contains a markdown table, it’s automatically rendered as an image.

Example:

You: Show me a comparison of programming languages

Daia: Here's a comparison:

| Language | Type | Use Case |
|----------|------|----------|
| Python | Interpreted | Data Science, Web |
| JavaScript | Interpreted | Web Development |
| Rust | Compiled | Systems Programming |

[Image of rendered table]

Supported Languages:

  • English
  • Simplified Chinese (简体中文)
  • Traditional Chinese (繁體中文)
  • Japanese (日本語)
  • Korean (한국어)

The rendering uses Noto Sans CJK fonts for proper display of all characters.

Smart Message Chunking

Discord has a 2000 character limit per message. Daia automatically splits long responses while preserving formatting:

  • Code blocks stay intact
  • Lists aren’t broken mid-item
  • Markdown formatting is preserved
  • Each chunk is sent as a separate message

Google Search Integration

Daia can automatically search the web when needed:

You: What's the latest version of Python?
Daia: [Searches Google] The latest version is Python 3.12...

This happens automatically when:

  • You ask about current events
  • You request up-to-date information
  • The bot determines a search would be helpful

User Recognition

With ENABLE_CONTEXTUAL_SYSTEM_PROMPT=on, Daia addresses users by their display name:

You: Hello!
Daia: Hi John! How can I help you today?

This creates a more personalized experience.

Advanced Configuration

Multiple Servers

You can use the same bot in multiple servers. Each server maintains its own conversation history per channel.

Channel-Specific Behavior

To restrict the bot to specific channels:

  1. Set ALLOWED_CHANNELS to only the channels where you want automatic responses
  2. In other channels, users must mention the bot
  3. Use Discord’s permission system to prevent the bot from seeing certain channels

Rate Limiting

Be aware of API rate limits:

  • Discord: 50 requests per second per bot
  • Gemini: Varies by tier (check your quota in Google AI Studio)

If you hit rate limits:

  • Reduce HISTORY_LIMIT to send fewer tokens per request
  • Implement cooldowns between messages (requires code modification)
  • Upgrade your Gemini API tier

Performance Tuning

For better performance:

  • Use gemini-1.5-flash instead of gemini-1.5-pro
  • Lower HISTORY_LIMIT to reduce context size
  • Reduce CHAT_TEMPERATURE for more consistent responses

Common Use Cases

Community Support Bot 

ALLOWED_CHANNELS=support-channel-id
ENABLE_CONTEXTUAL_SYSTEM_PROMPT=on
HISTORY_LIMIT=30

System prompt:

You are a support assistant for [Your Community].
Help users with common questions about [your topic].
Be patient and friendly. If you don't know something, direct users to the documentation.

Personal Assistant 

ALLOWED_CHANNELS=  # Leave empty, require mentions
ENABLE_CONTEXTUAL_SYSTEM_PROMPT=on
HISTORY_LIMIT=50

System prompt:

You are a personal AI assistant.
Help with research, brainstorming, and answering questions.
Provide detailed, thoughtful responses.

Multilingual Server 

ENABLE_CONTEXTUAL_SYSTEM_PROMPT=on
HISTORY_LIMIT=40

System prompt:

You are a multilingual AI assistant.
Respond in the same language the user uses.
Support English, Chinese, Japanese, and Korean.

Troubleshooting

Bot responds to every message

Issue: Bot is too chatty in channels.

Solution: Remove the channel from ALLOWED_CHANNELS so users must mention the bot.

Bot doesn’t remember context

Issue: Bot seems to forget previous messages.

Solution: Increase HISTORY_LIMIT in your .env file.

Responses are too random/inconsistent

Issue: Bot gives different answers to the same question.

Solution: Lower CHAT_TEMPERATURE (try 0.5 or 0.7).

Tables not rendering

Issue: Tables appear as text instead of images.

Solution:

  • Check that fonts were downloaded (check logs on first run)
  • Verify the bot has “Attach Files” permission
  • Check for errors in the console output

Next Steps