Securing coding agents and their code | BRK281
Ricky Diep and Rohan Parakh discuss the security controls needed to trust coding agents and their output, focusing on GitHub’s approach to agent-driven development.
Securing agents and AI-generated code
Coding agents can produce code in minutes, but organizations need confidence in both the agent’s execution environment and the code it generates. This session addresses three areas:
- Isolated sandboxes: Securing the environments in which coding agents operate.
- Supply chain protections: Protecting agent-driven development against supply chain threats.
- Automated code validation: Checking AI-generated code before deployment.
The discussion connects these controls to the broader challenge of expanding coding-agent use while maintaining security and governance across the development lifecycle.
Session details and resources
- Event: Microsoft Ignite 2026
- Session: BRK281 | Agent Platform
- Format and level: Breakout | (300) Advanced
- Speakers: Ricky Diep and Rohan Parakh
Explore the Agent Platform next-step resources.