Show HN: Coherence – drift detector for AI-driven repos (github.com)

Coherence, a new Git-native drift detector for AI-assisted repositories, has been released to help maintain consistency in AI-edited codebases. Unlike traditional AI reviewers, Coherence focuses on identifying discrepancies in repository components such as code, documentation, architectural decision records (ADRs), tests, and metrics—even after edits made by AI agents. This tool is designed to run locally and emphasizes data privacy, with optional LLM analysis that is disabled by default, ensuring that code does not leave the developer's machine. The significance of Coherence lies in its ability to pinpoint traceability and semantic consistency across a broader scope of repo components than existing tools. Traditional drift detectors often focus narrowly on structural changes, while Coherence evaluates the entire repository graph, offering improved detection of inconsistencies that could lead to bugs or broken functionality. This proactive approach enables developers to conduct thorough reviews before committing changes, thus minimizing future complications. Integrated with CI/CD workflows, Coherence can be used to enforce strict drift gating, enhancing the overall reliability of AI-enhanced software development.