Hadi2525/langgraph-builder

LangGraph Builder Agent is a modular Python project designed to facilitate the creation, management, and utilization of language graph-based agents. It provides tools for vector storage, agent orchestration, and integration with external resources, making it suitable for building advanced AI-driven applications.

• LangGraph Builder Agent
  • Table of Contents
  • Features
  • Project Structure
  • Installation
  • Usage
    • PDF Ingestion
  • Configuration
  • Vector Store
  • Extending the Agent
  • License

• Modular agent architecture
• Vector store integration (ChromaDB)
• PDF data ingestion
• Utility functions for agent workflows
• Configurable via JSON and YAML

agent/
  __init__.py         # Package initialization
  hub.py              # Agent hub and orchestration logic
  langgraph.json      # Agent configuration (JSON)
  main.py             # Entry point for running the agent
  spec.yml            # Agent specification (YAML)
  stub.py             # Agent stub/boilerplate
  utils.py            # Utility functions
  VectorStore.py      # Vector store abstraction
  vectorstore/        # Vector store data (ChromaDB)
data/
  pdf/                # PDF files for ingestion
vectorstore/          # ChromaDB vector store files
requirements.txt      # Python dependencies
README.md             # Project documentation
LICENSE               # License file

1. Clone the repository:

   git clone <repo-url>
   cd langgraph-builder

2. Install dependencies:

   pip install -r requirements.txt

Install the inmem for the langgraph-cli to make sure you installed all the dependencies:

pip install -U "langgraph-cli[inmem]"

Set the current directory to the agent folder:

cd agent

Run the agent inside LangGraph Studio (locally):

langgraph dev

Set the required values for the input json in the studio tool.

Place your PDF files in the data/pdf/ directory. The agent will process and index them into the vector store for retrieval and search.

I used the following book for building the vector store: Using Asyncio in Python.

• agent/langgraph.json: Main configuration for setting up the LangGraph Studio.
• agent/spec.yml: YAML specification for agent structure

• Uses ChromaDB for efficient vector storage and retrieval
• Vector data is stored in agent/vectorstore/ and vectorstore/
• VectorStore.py provides an abstraction layer for vector operations

• Add new agent logic in agent/hub.py or as new modules in agent/
• Utility functions can be added to agent/utils.py
• Update configuration files as needed for new features

See LICENSE for details.