Show HN: Privalyse – Catching Security Leaks in AI-Assisted Codebases

1 point

6 months ago

I just released Privalyse (v0.1 Alpha). The problem I faced: AI models can generate code very fast, but security and privacy context often gets lost — resulting in hardcoded API keys, PII ending up in logs, or unsafe input reaching sensitive sinks. Although reviews help, it’s easy to miss these issues when the PR volume is high.

Privalyse is a static analysis tool with cross-file taint tracking designed to catch privacy violations and security flaws that slip through unnoticed in AI-assisted workflows.

I'd love for you to give it a try! Since this is an early alpha, I'm specifically looking for: - Feedback on the detection rules - Reports on false positive patterns - Feature requests: What languages or frameworks should I prioritize next?

1 comment