Preview released: Detecting and fixing software vulnerabilities with CodeMender

As adversarial AI threats attack code at increasing speed, security teams need machine-speed defense systems with automated code remediation capabilities. CodeMender is a managed code security agent that provides the ability to scan software for vulnerabilities and fix them directly.

CodeMender can be delivered through the Gemini Enterprise Agent Platform or deployed as a core component of AI Threat Defense. It also supports a multi-model approach that lets users select the optimal model based on cost, speed, and deep-scan performance, with support for third-party frontier models planned by the end of this year.

Key capabilities include:

• Optimal model deployment: Choose from various models tailored to cost, speed, and deep-scan performance
• Machine-scale automated remediation: Eliminates bottlenecks from manual validation and patching while maintaining the development process
• Vulnerability prioritization: Verifies real-world exploitability by running and simulating PoC (Proof-of-Concept) exploits, addressing the most critical issues first

Built on research from Google DeepMind, CodeMender transforms vulnerability management from a manual task into an autonomous, high-speed system. The agent can integrate with existing CI/CD workflows or run directly in local development environments via a lightweight CLI client, while complying with enterprise-grade security guardrails such as secure routing through VPC and data isolation.