I’m not a fan of Deno’s security model
Summary
The post critiques Deno's security model, arguing that its permission-based approach is insecure and misleading. It examines prompts, terminal interactions, and environment-based attack vectors (PATH, LD_PRELOAD) to illustrate why Deno is not truly sandboxed. The author suggests that instead of relying on this pseudo-sandbox, a robust OS-level sandbox (like bubblewrap) is preferable and warns against harassing developers during discourse.