Lexiden - Legal Document Assistant

A conversational AI interface for generating legal documents using SSE streaming, LLM function calling, and prompt engineering.

Quick Start

Prerequisites

• Python 3.10+
• Node.js 18+
• OpenAI API Key

Backend Setup

cd backend
python -m venv venv
source venv/bin/activate  # Windows: venv\Scripts\activate
pip install -r requirements.txt
echo "OPENAI_API_KEY=sk-your-key-here" > .env
python app.py

Backend runs on http://localhost:5001

Frontend Setup

cd frontend
npm install
npm start

Frontend runs on http://localhost:3000

---

Project Structure & File Explanations

lexiden-challenge/
├── backend/
│   ├── app.py          # Flask server - SSE streaming & API endpoints
│   ├── prompts.py      # System prompt with engineering documentation
│   ├── tools.py        # Function definitions for LLM tool calling
│   └── requirements.txt
├── frontend/
│   └── src/
│       ├── App.js                    # Main layout with header & panels
│       ├── App.css                   # Global styles & design system
│       └── components/
│           ├── ChatInterface.js/css  # Chat UI with SSE handling
│           ├── MessageBubble.js/css  # Message display + tool indicators
│           └── DocumentPreview.js/css # Document viewer + edit history
└── README.md

Backend Files

app.py - Flask Server (Main Entry Point)

• Lines 37-53: Session management functions for conversation and document storage
• Lines 56-139: execute_function_call() - Handles the three LLM functions
• Lines 142-311: generate_sse_stream() - Core SSE streaming logic with OpenAI
• Lines 314-378: API endpoints (/api/chat, /api/conversation, /api/document, /api/health)

prompts.py - System Prompt

Contains the carefully engineered system prompt that defines:

• Conversation phases (Intent → Gathering → Generation → Revision)
• Required information per document type
• Function calling rules
• Edge case handling instructions

tools.py - Function Definitions

Defines three OpenAI-compatible function schemas:

1. extract_information - Structures user data
2. generate_document - Creates the document
3. apply_edits - Modifies existing documents

Frontend Files

App.js - Main Application

• State management for document and edit history
• Session ID generation for conversation tracking
• Layout with header, chat panel, and document panel

ChatInterface.js - Chat Component

• Lines 27-207: handleSendMessage() - Sends message and processes SSE stream
• Lines 74-188: SSE parsing loop that handles content, tool_call, tool_result, done, error events
• Real-time message updates as tokens stream in

MessageBubble.js - Message Display

• Renders user/assistant messages
• Shows tool call indicators with status (Processing/Complete)
• Displays streaming cursor animation

DocumentPreview.js - Document Viewer

• Renders document from structured data
• Highlights recently edited sections
• Shows edit history panel with version tracking

API Reference

Method	Endpoint	Description
POST	/api/chat	SSE streaming chat
GET	/api/conversation/:id	Get conversation history
DELETE	/api/conversation/:id	Clear conversation
GET	/api/document/:id	Get current document
GET	/api/health	Health check