AudiAI

AudiAI is an experimental, open-source technical auditing assistant that helps developers review authorized projects and understand potential security issues. The project combines AI agents and tools to work with project context, organize findings, and provide explanations and recommendations for human review. Its goal is to connect a potential issue with supporting evidence and a practical next step, rather than provide a generic AI response. AudiAI is read-only: it does not automatically modify the original project files. Findings and any proposed corrections require human verification and appropriate testing. For this hackathon, I adapted an existing agent-project foundation to develop the auditing workflow demonstrated in the video. The submission distinguishes the existing foundation from the work developed during the event. This is version 0.1, not a production-ready security product. It has known limitations and may produce incorrect or incomplete results. It does not replace professional security assessment or guarantee that a project is secure. The source code is published under GPL-3.0 so that the community can study it, report issues, and contribute improvements.