LangGraph Engineer LangGraph Engineer is an automated software development pipeline that leverages a multi-agent system to transform high-level user requirements into functional code. By utilizing a specialized Planner-Architect-Coder workflow, the system ensures that complex requests are broken down into logical engineering tasks before implementation begins.
🤖 The Agent Trio The system orchestrates three distinct personas to manage the development lifecycle:
The Planner: Analyzes the user's initial prompt to define the project scope, tech stack, and necessary file structure.
The Architect: Takes the high-level plan and decomposes it into a sequence of granular, dependency-aware implementation tasks.
The Coder: An iterative agent that executes the implementation tasks one by one, using ReAct logic to read and write files within the project directory.
🛠 Features Structured State Management: Uses Pydantic models to ensure strict data validation during agent hand-offs.
Safety Sandboxing: All file operations are restricted to a generated_project/ directory to prevent unauthorized system access.
Iterative Implementation: The Coder agent maintains a stateful index, allowing it to work through complex projects step-by-step rather than all at once.
Tool Integration: Includes built-in tools for reading/writing files, listing directories, and running shell commands.
🚀 Getting Started Prerequisites Python 3.9+
A Groq API Key (configured in your .env file)
Installation Clone the repository:
Bash
git clone https://github.com/Tammyb2/langgraph-engineer.git cd langgraph-engineer Install dependencies:
Bash
pip install langgraph langchain-groq pydantic python-dotenv Usage Run the main entry point and provide a prompt when requested:
Bash
python main.py Example prompt: "Create a Todo list app."
📂 Project Structure graph.py: Defines the LangGraph state machine and agent nodes.
states.py: Contains Pydantic schemas for project plans and agent states.
tools.py: Provides the filesystem and shell tools used by the Coder.
prompts.py: Houses the specialized system instructions for each agent.