Skip to content

README.md

Latest commit

 

History

History
153 lines (109 loc) · 3.67 KB

README.md

File metadata and controls

153 lines (109 loc) · 3.67 KB

Dynamic Memory System

A full-stack chat application built with FastAPI, Google ADK (Agent Development Kit), React, and DaisyUI.

Remembers User persona across various conversations and builds a timeline of user activities. This is achieved using context engineering techniques like entity extraction, llm as orchestrator & state resolver, tool calling

Features

  • A ChatGPT style chat UI that has Real-time streaming chat responses via SSE
  • Conversation history with auto-generated titles
  • Responsive UI with sidebar navigation
  • Supabase Integration for persistent storage of users and chat sessions
  • Mongodb Integration for storing user's persona and activity timeline
  • JWT-based User authentication

The moat of this project can be found at https://github.com/AdityaSanthosh/UserMemory-Chat/blob/main/backend/agent/agent.py

Tech Stack

Backend

  • FastAPI - Modern Python web framework
  • Google ADK - Agent Development Kit for managing AI agents
  • Supabase - PostgreSQL database for users and sessions

Frontend

  • React 18 - UI framework
  • React Router - Client-side routing
  • DaisyUI 4 - Tailwind CSS component library
  • Vite - Build tool

Setup

Prerequisites

  • Python 3.10+
  • Node.js 18+
  • Google Cloud API Key with Gemini API access
  • Supabase Account and Project

Supabase Setup

Create the following tables in your Supabase project:

  1. users

    • id (uuid, primary key)
    • username (text, unique)
    • hashed_password (text)
    • created_at (timestamptz)

  2. sessions

    • id (text, primary key)
    • app_name (text)
    • user_id (text)
    • state (jsonb)
    • last_update_time (timestamptz)

  3. events

    • id (int8/uuid, primary key, auto-generated)
    • session_id (text, foreign key to sessions.id)
    • event_data (jsonb)
    • created_at (timestamptz)

Environment Variables

Copy .env.example to .env and fill in the values:

cp .env.example .env

Required variables:

  • GOOGLE_API_KEY - Your Google API key for Gemini
  • SECRET_KEY - Secret key for JWT tokens (generate a random string)
  • SUPABASE_URL - Your Supabase project URL
  • SUPABASE_KEY - Your Supabase anon public key

Backend Setup

# Install Python dependencies
pip install -r requirements.txt

# Run the backend server
uvicorn backend.main:app --reload --port 8000

Frontend Setup

# Navigate to frontend directory
cd frontend

# Install dependencies
npm install

# Run development server
npm run dev

The frontend will be available at http://localhost:5173 and will proxy API requests to the backend at http://localhost:8000.

API Endpoints

Authentication

  • POST /api/auth/register - Create new user
  • POST /api/auth/login - Login and get JWT token
  • GET /api/auth/me - Get current user info

Sessions

  • GET /api/sessions - List user's chat sessions
  • GET /api/sessions/{id} - Get session with messages

Chat

  • POST /api/chat - Send message and receive streaming response (SSE)

SSE Events

The chat endpoint streams responses using Server-Sent Events:

  • session - {session_id, is_new} - Session info
  • delta - {content} - Streamed text chunk
  • title - {title} - Auto-generated title (for new sessions)
  • done - {} - Stream complete
  • error - {message} - Error occurred

Development

Running Both Servers

In one terminal:

uvicorn backend.main:app --reload --port 8000

In another terminal:

cd frontend && npm run dev

Building for Production

# Build frontend
cd frontend && npm run build

# The built files will be in frontend/dist
# The FastAPI app can serve these static files

License

MIT