Analyze Discord message history to understand trends and classify conversations with AI.

We built DinoDetect to analyze Discord conversations, especially for student org servers, by scraping message history and using AI to classify and summarize trends.

DinoDetect is a tool for scraping and processing Discord messages to uncover trends and classify message history in a meaningful way.
It combines JavaScript/Node.js, Discord’s API for message retrieval, and OpenAI’s API for analysis and categorization.

Make sure you have:

• Node.js (LTS recommended)
• A Discord Bot token / API access
• An OpenAI API key

Install packages:

npm install

Create a .env file in the project root (or set these in your shell):

touch .env

Add your keys (names may vary—adjust to your code):

DISCORD_BOT_TOKEN="your_discord_token"
OPENAI_API_KEY="your_openai_key"
# Optional:
# GUILD_ID="..."
# CHANNEL_ID="..."

Start the app using whatever script your repo defines in package.json:

npm run start

If your project doesn’t define a start script yet, run the main entry file directly (adjust path/name):

node index.js

• Building an end-to-end pipeline from scraping → processing → analysis
• Seeing trends emerge from real message data
• Applying “signals” thinking in a new domain (chat analytics)
• Making a fun demo video

• Add a proper database layer (initial work started, time constrained)
• Expand AI analysis beyond trends (e.g., toxicity and other safety signals)
• Embed analysis visuals back into Discord (bots/embeds), and improve UX

• Raul Valle — GitHub
• Sebastian Sosa — GitHub
• Matheus Kunzler Maldaner — GitHub