J^2 - Agents for Trades

J^2 is a multi-agent trading and compliance system where autonomous agents represent different investment personas, debate on assets using diverse datasets, and automatically execute trades while ensuring compliance with organizational SOPs. It merges the ideas of agentic AI debate, quantitative backtesting, and compliance automation into a single pipeline.

Key Features

• Multi-Agent Architecture: Specialized agents for data ingestion, processing, analysis, evaluation, and execution
• Comprehensive Data Processing: Advanced PDF extraction with OCR, semantic search, and RAG capabilities
• Financial Intelligence: Real-time market data, SEC filings, news analysis, and financial calculations
• Shared Tools Library: 80+ production-ready functions across 13+ tool categories
• Enterprise Security: Sandboxed code execution, encryption, and comprehensive logging

[📋] Table of Contents

Section	Description
Contributors	Meet the team behind this project
Tech Stack	Technologies, frameworks, APIs used in development
Project Structure	Organization of the codebase and purposes of different directories
Getting Started	Local setup instructions and prerequisites

[🧑‍💻] Contributors

Name	Role	GitHub
Jason Matthew Suhari	Full Stack, AI/ML, Agent Architecture	@jasonsuhari
Neo Jing Yi	Co-Developer, Data Engineering	@neojingyi

[⚙️] Tech Stack

Frontend

Backend

Databases & Storage

AI/LLM & Machine Learning

Data Processing & Analysis

Visualization & Reporting

Search & Information Retrieval

Communication & Notifications

Development & Utilities

[📁] Project Structure

first-visions-danta/
├──  frontend/                    # Next.js frontend application
│   ├── app/                       # App Router pages & layouts
│   ├── components/                # Reusable UI components
│   │   ├── threads.tsx           # Thread management UI
│   │   └── rotating-text.tsx     # Dynamic text animations
│   ├── lib/                      # Frontend utilities
│   └── public/                   # Static assets
│
├──  agents/                     # Multi-Agent System
│   ├── shared_tools/             # Comprehensive shared tools library (80+ functions)
│   │   ├── web_search.py         # SerpAPI, NewsAPI, SEC Edgar search
│   │   ├── vector_search.py      # ChromaDB, embeddings, semantic search
│   │   ├── document_rag.py       # PyMuPDF → OCR PDF processing + RAG
│   │   ├── email_service.py      # SMTP, SendGrid, templates
│   │   ├── notification_service.py # Slack, Teams, Telegram, SMS
│   │   ├── chat_interface.py     # OpenAI/Anthropic chat with context
│   │   ├── code_executor.py      # Sandboxed Python/R/SQL execution
│   │   ├── math_calculator.py    # Financial calculations & statistics
│   │   ├── data_processor.py     # CSV/JSON/XML processing
│   │   ├── chart_generator.py    # Matplotlib/Plotly visualizations
│   │   ├── report_builder.py     # PDF/HTML report generation
│   │   ├── table_formatter.py    # Financial table styling
│   │   ├── file_manager.py       # Local & cloud storage management
│   │   ├── test_all_tools.py     # Comprehensive test suite
│   │   └── requirements.txt      # All dependencies
│   │
│   ├── 01_data_ingestion_agents/  # Data collection & processing
│   │   └── macro_data/           # Economic data ingestion
│   ├── 02_data_processing_agents/ # Data transformation & analysis
│   │   └── nl_normalizer/        # Natural language data normalizer
│   ├── 03_analysis_agents/       # Financial analysis & insights
│   ├── 04_execution_agents/      # Trade execution & automation
│   └── 05_evaluation_agents/     # Performance evaluation
│       └── performance_evaluator/ # Trading performance analysis
│
├──  backend/                   # FastAPI backend services
│   ├── apps/                     # Application modules
│   │   ├── shared_utils/         # Backend utilities & connectors
│   │   └── main.py              # FastAPI entry point
│   └── tests/                    # Backend test suite
│
├──  data/                      # Data storage & processing
├──  scripts/                   # Utility scripts
└──  docs/                      # Documentation & guides

[🔰] Getting Started

Prerequisites

• Conda/Miniconda: Install here
• Node.js 18+: Install here
• Tesseract OCR: For PDF text extraction (Windows, macOS: brew install tesseract, Ubuntu: sudo apt install tesseract-ocr)
• Git: For cloning the repository

Installation

1. Clone the repository

   git clone https://github.com/jasonsuhari/first-visions-danta.git
   cd first-visions-danta

2. Setup Shared Tools (Core agent capabilities)

   cd agents/shared_tools
   pip install -r requirements.txt
   
   # Copy environment template and add your API keys
   cp env_template.txt .env
   # Edit .env with your API keys (OpenAI, SerpAPI, etc.)
   
   # Test the installation
   python test_all_tools.py

3. Setup Frontend

   cd frontend
   npm install
   npm run dev

4. Setup Backend (in a separate terminal)

   cd backend
   bash setup.bash  # Creates conda environment and installs dependencies
   conda activate first-visions-danta-env
   python main.py

Required API Keys

For full functionality, add these to your .env file:

• Search: SERP_API_KEY, NEWS_API_KEY
• AI/LLM: OPENAI_API_KEY
• Email: GMAIL_EMAIL, GMAIL_APP_PASSWORD
• Notifications: DISCORD_WEBHOOK_URL
• Database: SUPABASE_URL, SUPABASE_KEY

What You Get

After setup, you'll have access to:

• 80+ Production-Ready Functions across 13 tool categories
• Advanced PDF Processing with PyMuPDF → OCR fallback
• Multi-Agent System for financial analysis and trading
• Real-time Data Processing with market data integration
• Beautiful UI with interactive charts and dashboards

---

Built with ❤️ by the J^2 Team.