Vibe Coding with AI is Cool Until You Get Hacked

There are a lot of new advancements coming up in the AI landscape like Vibe Coding, MCP, Agents2Agents. Not just that, people are using AI coding assistant tools in their day-to-day life. There is more hype around AI, but wait we tend to forget one thing, and that is security.

There are a lot of problems in consuming all those AI coding assistant tools and LLMs. How do you know the code generated by your favourite coding assistant is not vulnerable? How do you verify it? Your code might use deprecated libraries or vulnerable ones too, right? LLMs have their own knowledge cut-off date, right? Then how do you make sure you are writing secure and clean code?

Don’t worry that’s where Codegate comes in, one of the popular open-source projects which solves these problems. Codegate acts as a proxy between your LLMs and coding assistants, analyzes the code generated by LLMs, fixes it, and recommends best practices to write secure code. All things happen locally nothing goes outside of your machine.

Most people are not aware of this popular open-source project. In my talk, I will show how Codegate helps you to consume the code generated by your favourite coding tools.

Key Takeaways:

• Understand the real risks of relying blindly on AI coding assistants and what you can do about it.

• See how Codegate can act like your personal security reviewer for AI-generated code.

• Learn how to spot things like deprecated libraries or hidden vulnerabilities before they hit production.

• Walk away with some practical security best practices you can start using right away.

• Discover some cool open-source projects in the AI + security space that most people haven’t even heard of.