Discord Analysis Bot Example

This example demonstrates how to build a Discord analysis bot that categorizes and classifies messages from a help forum to minimize the time needed to understand customer issues.

Overview

The Discord Analysis Bot:

1. Scrapes messages from the help-and-questions forum in Discord
2. Uses an agent to categorize messages based on Mastra concepts (Workflows, Memory, Primitives, Docs, Getting Started)
3. Generates a structured analysis with categories, counts, sentiment, and top issues

Prerequisites

• Node.js v20.0+
• pnpm (recommended) or npm
• OpenAI API key
• Discord bot token (optional, for real Discord API access)

Getting Started

1. Clone the repository and navigate to the project directory:

   git clone https://github.com/mastra-ai/mastra
   cd examples/discord-analysis-bot

2. Install dependencies:

   pnpm install

3. Create a .env file with your API keys:

   OPENAI_API_KEY=your_openai_api_key
   DISCORD_BOT_TOKEN=your_discord_bot_token  # Optional
   DISCORD_CHANNEL_ID=your_channel_id        # Optional
   

4. Run the example:

   pnpm start

Setting Up a Discord Bot (Optional)

If you want to use real Discord data instead of mock data:

1. Create a Discord application:

   • Go to Discord Developer Portal
   • Click "New Application" and give it a name
   • Navigate to the "Bot" tab and click "Add Bot"

2. Configure bot permissions:

   • Under "Privileged Gateway Intents", enable "Message Content Intent"
   • This allows the bot to read message content

3. Get your bot token:

   • In the Bot tab, click "Reset Token" or "Copy" to get your bot token
   • Add this token to your .env file as DISCORD_BOT_TOKEN

4. Invite the bot to your server:

   • Go to the "OAuth2" tab, then "URL Generator"
   • Select "bot" under scopes
   • Select permissions: "Read Messages/View Channels", "Read Message History"
   • Copy the generated URL and open it in your browser
   • Select your server and authorize the bot

5. Get your channel ID:

   • Enable Developer Mode in Discord (Settings > Advanced > Developer Mode)
   • Right-click on the channel you want to analyze and select "Copy ID"
   • Add this ID to your .env file as DISCORD_CHANNEL_ID

Project Structure

• src/index.ts: Main entry point that runs the Discord analysis
• src/mastra/index.ts: Initializes the Mastra instance
• src/mastra/agents/index.ts: Defines the Discord analysis agent
• src/mastra/tools/index.ts: Contains the Discord scraper tool

Example Output

The example will output:

• Categories of messages (Workflows, Memory, Primitives, etc.)
• Count of messages in each category
• Sentiment analysis for each category
• Top issues identified in each category
• Overall summary of the analysis

Implementation Notes

This example demonstrates how to build a Discord analysis bot that connects to Discord and fetches actual messages for analysis. The implementation:

1. Uses the Discord.js library to connect to Discord and retrieve messages
2. Implements proper authentication with a Discord bot token
3. Includes error handling and logging for API calls
4. Provides structured analysis of message categories and issues

For a production implementation, you might consider adding features like:

• Scheduled analysis runs
• Historical data storage
• Trend analysis over time
• Integration with notification systems