Publication
Cindy: A Toolkit for Automated Architecture Recovery in Large Software Repositories
Martin Marko; Lukas Radosky; Tomas Uhercik; André Antakli; Ivan Polasek
In: 2026 6th International Conference on Electrical, Computer and Energy Technologies (ICECET). International Conference on Electrical, Computer and Energy Technologies (ICECET-2026), July 6-9, Rome, Italy, Pages 1-6, IEEE, 2026.
Abstract
Large language models (LLMs) and AI coding agents are increasingly used in software development workflows. However, their effectiveness degrades significantly when applied to large legacy systems. This paper presents the Cindy Toolkit, a comprehensive suite of tools designed to facilitate deep, multi-faceted repository comprehension by analyzing not just the source code but also the entire ecosystem of artifacts within a repository. We demonstrate how just a small subset of tools contained in Cindy can be used to generate comprehensive human-readable reports on various aspects of software. We also suggest possibilities of integration into AI-driven workflows to provide agents with the contextual understanding necessary for successful task execution.
