IDE Plugin for Repository Assistant

Introduction

The IDE Plugin for Repository Assistant is a key component of the broader Repository Assistant project, which employs Large Language Models (LLMs) to enhance software development processes. This project extends the tool's functionality to provide more precise, context-sensitive support directly within developers' Integrated Development Environments (IDEs) like IntelliJ and VSCode.

Motivated by the challenges developers face with code comprehension and generation, including issues like model hallucinations, security concerns, and efficiency, this project aims to refine the interaction between LLMs and the coding environment. By gathering contextual information from the IDE, the Repository Assistant will offer enhanced question-and-answer support, helping developers efficiently navigate and understand their code bases.

Task Overview

The master project focuses on developing a plugin that improves various components of the Repository Assistant tool. Specific tasks include:

• Gathering relevant information from the development environment to generate potential answers to developers' questions about a piece of code.
• Building the tool interface within IDEs like VS Code or IntelliJ to seamlessly integrate these capabilities.
• Ensuring that the model can be trained securely and locally, addressing privacy and data security concerns effectively.

Technical Skills and Technologies:

This project requires software development skills, including proficiency in Python, PyTorch, and the ability to work with the GitHub API, HTML/CSS/JavaScript, and REST APIs. Knowledge of SQL/NoSQL databases and containerization with Docker is also important to manage backend processes. Additionally, experience or a keen willingness to work with Large Language Models, such as CodeLlama, is essential.

Contact

For further details and inquiries about this project, please contact:

• Alex Wolf: wolf@ifi.uzh.ch
• Marco Palma: marcoepalma@ifi.uzh.ch
• Dr. Pooja Rani: rani@ifi.uzh.ch