AI-Powered GitHub Repository Reviewer and Developer Assistant: A Literature Review and Conceptual Framework
In this paper, there is going to be a literature review and a conceptual framework created for the AI-powered GitHub Repository Reviewer and Developer Assistant. In this work, it will be discussed how the use of LLMs, RAG, code embeddings, and AI agents can be beneficial for the analysis of large software repositories by developers. There is a review of literature related to automatic code review, repository-level bugs detection, AI-driven software engineering agents, code retrieval methods. Also, key problems have been identified, such as insufficient repository context, possible errors of the code retrieval, wrong AI-generated advice, security and privacy risks. The conceptual framework includes repository processing, code embeddings, vector database, code retrieval with the use of RAG, analysis with LLMs, and AI agents. The outcome of the framework will be context-aware code reviews and recommendations. It has been mentioned before, that there is a conceptual framework created with the use of the literature review. There is no implementation of the framework or its evaluation in experiments.
Authors
- Bhumika Pardeshi
Institutions
- Ajeenkya DY Patil University (IN)
Publication Details
- Journal
- Zenodo (CERN European Organization for Nuclear Research)
- Published
- 2026-11-19
- DOI
- https://doi.org/10.5281/zenodo.24372614
- Primary Topic
- Software Engineering Research
- Type
- article
- Field-Weighted Citation Impact
- 0.00