AI coding assistants are often presented as tools to write code faster. But is that code vulnerable? Or is it safe by default? Can we actually make it comply with company standards and security best practices?

We have plenty of techniques—agents, skills, prompts and hooks—that can help us move toward a security‑by‑default approach. Our proposal is to leverage these capabilities to identify and fix security weaknesses in an existing application by analysing code, detecting insecure patterns, and strengthening configurations.

Starting from a vulnerable application, and using an AI‑assisted security approach, the objective will be to harden the application through workflows that analyse code, detect insecure patterns, review Dockerfiles, strengthen configurations, and improve pipelines.

Finally, the goal is to showcase how this approach can help teams significantly improve the security posture of their applications and development workflows. In other words, deploy AI-generated code with confidence.