Secure agent workflows in GitHub Copilot with NVIDIA OpenShell | DEMSP387
Key Takeaways
Secure agent workflows in GitHub Copilot with NVIDIA OpenShell
Original Description
Agentic workflows in GitHub Copilot can run complex tasks with broad access to local systems. In this session, see how to enforce policy and governance for agents operating in CLI environments. You’ll walk through how NVIDIA OpenShell applies kernel-level controls, isolates agent processes, and limits permissions based on task context. Learn how to apply zero-trust principles to secure agent-driven development workflows.
Seating for this session is first-come, first-served. Add it to your schedule to plan your day and arrive early to secure a spot.
𝗦𝗽𝗲𝗮𝗸𝗲𝗿𝘀:
* Ali Golshan
* Alex Watson
𝗦𝗲𝘀𝘀𝗶𝗼𝗻 𝗜𝗻𝗳𝗼𝗿𝗺𝗮𝘁𝗶𝗼𝗻:
This is one of many sessions from the Microsoft Build 2026 event. View even more sessions on-demand and learn about Microsoft Build at https://build.microsoft.com
DEMSP387 | English (US) | Developer tools & frameworks
Demo | (200) Intermediate
#MSBuild
Chapters:
0:00 - Deterministic policy and governance at infrastructure level
00:03:43 - Core runtime primitives: gateway, sandboxes, and policy enforcement
00:06:09 - Introduction of unique capability through policy prover technology
00:08:23 - Three privacy router use cases for managing PII and inference routing
00:11:16 - Policy enforcement and automation introduction
00:12:00 - Sandbox restart and concurrency capabilities
00:14:24 - Final results written to GitHub and sandbox cleanup
00:18:11 - Overview of Open Shell ecosystem and partnerships
00:21:07 - Explanation of dynamic agent security policy negotiation under zero trust model
Watch on YouTube ↗
(saves to browser)
Sign in to unlock AI tutor explanation · ⚡30
More on: Agent Foundations
View skill →Related AI Lessons
⚡
⚡
⚡
⚡
My agent kept reading data it wasn't allowed to. The prompt was never going to stop it.
Dev.to AI
8 Must-Know AI Chatbot Tools That Actually Help Small Businesses
Dev.to AI
Agent-Ready Commerce, Part 9: Evidence and Audit Are Part of the Product
Dev.to AI
Agent-Ready Commerce, Part 8: Generated Claims Need Review, Evidence, and Expiry
Dev.to AI
Chapters (9)
Deterministic policy and governance at infrastructure level
3:43
Core runtime primitives: gateway, sandboxes, and policy enforcement
6:09
Introduction of unique capability through policy prover technology
8:23
Three privacy router use cases for managing PII and inference routing
11:16
Policy enforcement and automation introduction
12:00
Sandbox restart and concurrency capabilities
14:24
Final results written to GitHub and sandbox cleanup
18:11
Overview of Open Shell ecosystem and partnerships
21:07
Explanation of dynamic agent security policy negotiation under zero trust mode
🎓
Tutor Explanation
DeepCamp AI