Apiiro’s Guardian Agent guards against insecure AI code

Apiiro has launched Guardian Agent, an AI agent that helps prevent coding agents from generating vulnerable or non-compliant code by rewriting developer prompts into secure prompts, according to the company.

Introduced January 28, Guardian Agent is now in a private preview stage. Describing the technology as introducing a fundamentally new paradigm for securing software in the era of AI-driven development, Apiiro said Guardian replaces traditional appsec approaches built around detecting and fixing vulnerabilities after code is written. Guardian Agent replaces this reactive model with a preventive one, stopping risk before code is generated by guarding AI coding agents in real time, according to Apiiro. Guardian Agent operates in real time directly from the developer’s IDE and CLI tools. The agent is powered by Apiiro’s code analysis technology and a software graph that “deeply understands” the customer’s software architecture and adapts to its changes, the company said.

Elaborating on the inspiration behind Guardian Agent, Apiiro said AI coding agents are breaking the physics of application security. Enterprises generate four times more code after adopting AI coding agents and expand the application attack surface by six times. This expansion is driven by rapid generation of new APIs, duplicated open source technologies and dependencies, and other resources, reshaping the software architecture with each code change, Apiiro said. Much of the code is generated without developers being fully aware of it. By preventing vulnerabilities before code exists, security outcomes are improved and developer productivity is increased, Apiiro stressed.