CAS - Codebase Agent

This project is a Retrieval-Augmented Generation (RAG) agent designed to understand and interact with codebases. It ingests external repositories, indexes them using vector embeddings, and allows users to query the codebase using a powerful LLM.

Table of Contents

• Tech Stack
• Project Structure
• How to use?

Tech Stack used

Stack	Tech
Language
Frameworks
AI / ML
Database
Utilities

Project Structure

RAG/
│
├── add_repo.py                     # Script to ingest external repositories
├── run_app.bat                     # Batch script to run the application
├── requirements.txt                # Project dependencies
├── .env                            # Environment variables (API keys)
│
├── intern/                         # Core internal logic
│   ├── agent/                      # Agent definition
│   │   ├── graph.py                # LangGraph workflow definition
│   │   ├── state.py                # Agent state definition
│   │   └── tools.py                # Custom tools for the agent
│   │
│   ├── app/                        # FastAPI application
│   │   └── main.py                 # API entry point
│   │
│   ├── db/                         # Database handling
│   │
│   └── ingest/                     # Ingestion logic
│       └── ingest_codebase.py      # Logic to ingest code into ChromaDB
│
├── source_code/                    # Directory where ingested repos are stored
│   └── ...
│
├── testfiles/                      # Files for testing
│
└── venv/                           # Virtual environment

How to use?

1. Setup Environment

Ensure you have Python installed. Create a virtual environment and install dependencies.

python -m venv venv
venv\Scripts\activate
pip install -r requirements.txt

Create a .env file in the root directory and add your Google API key:

GOOGLE_API_KEY=your_api_key_here

2. Ingest a Repository

To add a codebase for the agent to understand, use the add_repo.py script. This will copy the repo to source_code/ and ingest it into the vector database.

python add_repo.py /path/to/external/repo

3. Run the Agent

You can start the FastAPI server using the provided batch script or directly with uvicorn.

Using Batch Script:

run_app.bat

Using Uvicorn:

cd intern
uvicorn app.main:app --reload

4. Query the Agent

Once the server is running (default: http://localhost:8000), you can send POST requests to interact with the agent.

Endpoint: POST /

Payload:

{
  "question": "Explain the main logic of the ingested codebase."
}

Author

Chirag Wattamwar