Discord Jarvis Bot 🤖

A powerful Discord bot that combines chat capabilities with document processing, vector search, and web search functionality. Jarvis can understand context from uploaded documents, search through chat history, and provide up-to-date information from the web.

✨ Features

• 💬 Conversational AI: Powered by OpenAI GPT or Open Source models with chat history context
• 📄 Document Processing: Automatically processes PDF, DOCX, and TXT file uploads
• 🔍 Vector Search: Semantic search through uploaded documents using TiDB Vector Search
• 🌐 Web Search: Real-time web search capabilities via Perplexity AI
• 📊 Full-Text Search: Search through chat history using TiDB's full-text search
• 🛠️ Tool Integration: Extensible tool system using FastMCP

🎥 Demo Video

https://www.loom.com/share/d117f2aaf329475999278ad3859c4460?sid=cde0add4-b238-4328-ad8b-6137a88db187

🏗️ Architecture

The bot uses several key components:

• Discord Bot: Main interface for user interactions
• TiDB Database: Stores chat history and attachments with vector search capabilities
• Vector Store: Semantic search through document chunks
• LLM Integration: OpenAI API for conversational responses
• MCP Tools: Modular tool system for extensible functionality

🚀 Quick Start

Prerequisites

• Python 3.8+
• TiDB Cloud account (free tier available)
• Discord Bot Token
• OpenAI API Key
• Perplexity AI API Key (optional, for web search)

Installation

1. Clone the repository

git clone https://github.com/ItshMoh/chat-jarvis

2. Install dependencies

pip install -r requirements.txt

3. Set up environment variables Create a .env file in the root directory:

# Discord Configuration
   DISCORD_BOT_TOKEN=your_discord_bot_token_here
   
   # TiDB Configuration
   TIDB_CONNECTION_URL=mysql+pymysql://username:password@host:port/database
   TIDB_SSL_CA=path/to/ca-cert.pem  # Optional, for SSL
   
   # OpenAI Configuration
   OPENAI_API_KEY=your_openai_api_key_here
   API_BASE_URL=https://api.openai.com/v1
   LLM_MODEL=gpt-4o
   TEMPERATURE=0.7
   MAX_TOKENS=1500
   
   # Perplexity AI (Optional - for web search)
   PPLX_API_KEY=your_perplexity_api_key_here
   
   # MCP Server Configuration
   MCP_HOST=127.0.0.1
   MCP_PORT=9096

Setting Up TiDB Cloud

• Create a TiDB Cloud account at tidbcloud.com
• Create a new cluster (Starter tier is free)
• Get connection details and update your .env file
• Enable Vector Search in your cluster settings

Setting Up Discord Bot

• Go to Discord Developer Portal at discord.com/developers/applications
• Create New Application and give it a name
• Go to Bot section and create a bot
• Copy the token and add it to your .env file
• Set permissions: Send Messages, Read Message History, Attach Files
• Invite bot to your server using the OAuth2 URL generator

🎮 Usage

Starting the Bot

1. Start the MCP server (in one terminal):

python tools.py

2. Start the Discord bot (in another terminal):

python main.py

Bot Commands

• /bot - Ask Jarvis anything

/bot What is machine learning?
/bot Summarize the document I just uploaded
/bot What's the latest news about AI?

File Upload Support

Simply upload supported files (PDF, DOCX, TXT) to any channel where Jarvis is present. The bot will automatically:

• Extract text content
• Create semantic chunks
• Store in vector database
• Make content searchable

🧠 How It Works

Document Processing Pipeline

• File Upload Detection → Automatic processing of attachments
• Text Extraction → Support for PDF, DOCX, TXT formats
• Chunking → Split into manageable pieces using RecursiveCharacterTextSplitter
• Embedding Generation → Create vector embeddings using sentence-transformers
• Vector Storage → Store in TiDB Vector for semantic search

Intelligent Response System

• Context Gathering → Retrieve chat history and relevant documents
• Tool Selection → Automatically choose appropriate tools (document search, web search, chat search)
• Response Generation → Generate contextual responses using OpenAI
• Memory Storage → Store conversations for future context

🛠️ Available Tools

These all tools are available to the bot with the help of MCP server.

1. get_context Searches through uploaded documents for relevant information

• Input: Query string, optional channel_id, optional author
• Output: Relevant document chunks with metadata

2. web_search Searches the web for current information using Perplexity AI

• Input: Search query
• Output: Current web information with sources

3. search_chats Searches through chat history using full-text search

• Input: List of keywords
• Output: Relevant chat messages

🔍 Troubleshooting

Common Issues

1. Bot not responding

• Check if both tools.py and main.py are running
• Verify Discord token and permissions
• Check console for error messages

2. Database connection errors

• Verify TiDB connection URL format
• Check network connectivity
• Ensure database exists and is accessible

3. File processing failures

• Check if required libraries are installed (PyPDF2, python-docx)
• Verify file size limits (50MB max)
• Check file format support

4. Vector search not working

• Ensure TiDB Vector is enabled in your cluster
• Check embedding dimension compatibility
• Verify vector table creation