Prompt2Code

This project converts natural language prompts into code using AI agents and tools.

Features

• Python backend for agent logic
• Web frontend in generated_project/ (HTML, CSS, JS)
• Modular design with agent/ for graph, prompt, states, and tools

Project Overview

Prompt2Code is an AI-powered engineering project generator. It takes natural language prompts and produces a complete codebase, including backend logic and a web frontend.

Architecture

• Agent System:
  • Modular agents: Planner, Architect, Coder
  • Uses LangChain, LangGraph, and Groq LLM for reasoning and code generation
  • Agents communicate via a state graph (agent/graph.py)
• Tools:
  • File operations, directory listing, and shell commands are abstracted in agent/tools.py
• State Management:
  • Project plans, implementation steps, and file purposes are modeled in agent/states.py
• Prompts:
  • Custom prompt templates for each agent in agent/prompt.py

Web Frontend

• Located in generated_project/
• Includes a responsive Todo App (index.html, styles.css, script.js)
• Features: Add, edit, delete, filter, and persist todos in local storage

Python Dependencies

• groq, langchain, langchain-core, langchain-groq, langgraph, pydantic, python-dotenv

Usage

Configuration

• Set your Groq API key in .env (never commit secrets)

IDE Support

• .idea/ folder for JetBrains IDEs (PyCharm, etc.)
• VS Code settings supported via .vscode/

Extending

• Add new agent logic in agent/
• Extend frontend features in generated_project/

Setup

1. Clone the repository
2. Install Python dependencies (see pyproject.toml)
3. Configure your environment variables in .env (do not commit secrets)
4. Start the backend: python main.py
5. Open generated_project/index.html in your browser

Folder Structure

• main.py — Entry point for backend
• agent/ — Core agent logic
• generated_project/ — Web frontend

Contributing

Pull requests are welcome. For major changes, please open an issue first.

License

MIT