Pydantic Chatbot

A full-stack multi-agent chat system with specialized AI agents for Python code generation, web search, and image analysis.

Features

• Multi-Agent System: Specialized AI agents for different tasks
  • Code Expert: Generates, explains, and executes Python code
  • Search Expert: Retrieves and summarizes information from the web
  • Image Expert: Analyzes uploaded images with object detection
• Dual Response Modes: Streaming and structured output
• Interactive UI: Session management, chat history, markdown support
• Additional: Image upload, Python code execution, web search with source attribution, mobile-responsive design

Architecture

• backend/ — FastAPI backend (see backend/README.md for setup)
• frontend/ — Next.js frontend (see frontend/README.md for setup)

Unified Development with Makefile

Use the Makefile for easy development:

• make dev — Run both backend and frontend (recommended for local development)
• make backend — Run only the backend API
• make frontend — Run only the frontend UI

Environment Variables

Create a .env file in the project root with:

GEMINI_API_KEY=your_gemini_api_key
OPENAI_API_KEY=your_openai_api_key (optional)
LOGFIRE_TOKEN=your_logfire_token_if_using_logfire

And a .env.local file in the frontend directory with:

NEXT_PUBLIC_API_BASE_URL=http://localhost:8000

Technology Stack

• Backend: FastAPI, Pydantic, SQLite
• Frontend: Next.js, TypeScript, Material UI
• AI: OpenAI GPT-4o, Google Gemini 2.5 Flash
• Tooling: Uvicorn, Logfire

More Information

• For backend setup, API docs, and advanced usage, see backend/README.md
• For frontend setup and development, see frontend/README.md