Nvidia OpenShell: Difference between revisions
From EverybodyWiki Bios & Wiki
Moved page from wikipedia |
m remove duplicates internal links |
||
| Line 22: | Line 22: | ||
}} | }} | ||
'''Nvidia OpenShell''' is an [[open-source software|open-source]] runtime developed by | '''Nvidia OpenShell''' is an [[open-source software|open-source]] runtime developed by Nvidia for executing autonomous [[AI agent]]s in isolated [[sandbox (computer security)|sandbox]] environments. It controls agents' access to [[Computer file|files]], [[Process (computing)|processes]], [[Computer network|networks]], external services, and [[Credential|credentials]] through [[Kernel (operating system)|kernel]]-level isolation and [[Declarative programming|declarative]] security policies.<ref name="overview">{{cite web |title=Overview of NVIDIA OpenShell |url=https://docs.nvidia.com/openshell/latest/about/why-open-shell |website=NVIDIA OpenShell Developer Guide |publisher=Nvidia |access-date=September 28, 2026}}</ref><ref name="canonical">{{cite web |title=Securing AI agent workflows on Ubuntu with the new NVIDIA OpenShell snap |url=https://canonical.com/blog/nvidia-openshell-ubuntu-announcement |website=Canonical |date=June 1, 2026 |access-date=September 28, 2026}}</ref> | ||
OpenShell was publicly previewed for [[Ubuntu]] at [[Computex]] in June 2026 as part of a collaboration between Nvidia and [[Canonical (company)|Canonical]].<ref name="canonical" /> Version 0.1.0 expanded the runtime with policy [[Formal verification|verification]], credential-protected service access, support for multi-tenant deployments, and [[Central processing unit|CPU]] and [[graphics processing unit|GPU]] execution.<ref name="nvidia">{{cite web |last1=Watson |first1=Alex |last2=Golshan |first2=Ali |title=Add Runtime Controls to AI Agents with NVIDIA OpenShell |url=https://developer.nvidia.com/blog/add-runtime-controls-to-ai-agents-with-nvidia-openshell/ |website=NVIDIA Technical Blog |publisher=Nvidia |date=September 28, 2026 |access-date=September 28, 2026}}</ref> | OpenShell was publicly previewed for [[Ubuntu]] at [[Computex]] in June 2026 as part of a collaboration between Nvidia and [[Canonical (company)|Canonical]].<ref name="canonical" /> Version 0.1.0 expanded the runtime with policy [[Formal verification|verification]], credential-protected service access, support for multi-tenant deployments, and [[Central processing unit|CPU]] and [[graphics processing unit|GPU]] execution.<ref name="nvidia">{{cite web |last1=Watson |first1=Alex |last2=Golshan |first2=Ali |title=Add Runtime Controls to AI Agents with NVIDIA OpenShell |url=https://developer.nvidia.com/blog/add-runtime-controls-to-ai-agents-with-nvidia-openshell/ |website=NVIDIA Technical Blog |publisher=Nvidia |date=September 28, 2026 |access-date=September 28, 2026}}</ref> | ||
== Architecture == | == Architecture == | ||
OpenShell consists primarily of a gateway, supervisor, and | OpenShell consists primarily of a gateway, supervisor, and sandbox. The gateway manages sandboxes and their policies, while the supervisor controls communication between an agent and external services.<ref>{{Cite web |date=2026-09-29 |title=Amid calls for ‘pacing,’ a new Nvidia safety tool to stop AI agents from going rogue |url=https://indianexpress.com/article/explained/explained-ai/nvidia-open-agent-safety-platform-openshell-sentry-10898569/ |access-date=2026-09-30 |website=The Indian Express |language=en}}</ref> Agent workloads execute inside sandboxes with operating-system controls restricting filesystem and process [[Computer access control|access]].<ref name="nvidia" /> | ||
Policies are defined declaratively and can restrict network destinations and individual operations on services. OpenShell can also keep | Policies are defined declaratively and can restrict network destinations and individual operations on services. OpenShell can also keep credentials outside the agent workload and provide them only to [[Authorization|authorized]] requests. Its policy prover uses [[formal methods]] to analyze whether proposed permissions exceed defined [[Trust boundary|security boundaries]].<ref name="nvidia" /> | ||
OpenShell supports local and remote execution using [[Docker (software)|Docker]], [[Podman]], [[Virtual machine|MicroVMs]], and [[Kubernetes]].<ref>{{cite web |title=Support Matrix |url=https://docs.nvidia.com/openshell/reference/support-matrix |website=NVIDIA OpenShell Developer Guide |publisher=Nvidia |access-date=September 28, 2026}}</ref><ref>{{Cite web |date=2026-09-28 |title=Nvidia says its new OpenShell platform can stop AI agents from going rogue - CBS News |url=https://www.cbsnews.com/news/nvidia-openshell-ai-safety-rogue-agents/ |access-date=2026-09-30 |website=www.cbsnews.com |language=en-US}}</ref> | OpenShell supports local and remote execution using [[Docker (software)|Docker]], [[Podman]], [[Virtual machine|MicroVMs]], and [[Kubernetes]].<ref>{{cite web |title=Support Matrix |url=https://docs.nvidia.com/openshell/reference/support-matrix |website=NVIDIA OpenShell Developer Guide |publisher=Nvidia |access-date=September 28, 2026}}</ref><ref>{{Cite web |date=2026-09-28 |title=Nvidia says its new OpenShell platform can stop AI agents from going rogue - CBS News |url=https://www.cbsnews.com/news/nvidia-openshell-ai-safety-rogue-agents/ |access-date=2026-09-30 |website=www.cbsnews.com |language=en-US}}</ref> | ||
Latest revision as of 02:16, 28 October 2026
Another editor has reviewed this page's proposed deletion, endorses the proposal to delete, and adds:
If you remove the {{proposed deletion/dated}} tag above, please also remove this {{Proposed deletion endorsed}} tag. |
 | |
| Developer(s) | Nvidia |
|---|---|
| Stable release | 0.1.2
|
| Repository | https://github.com/NVIDIA/OpenShell |
| Written in | Rust, Go, Shell |
| Engine | |
| Operating system | Linux, macOS, Windows (WSL 2) |
| Type | Artificial intelligence agent runtime |
| License | Apache License 2.0 |
| Website | docs |
Search Nvidia OpenShell on Amazon.
Nvidia OpenShell is an open-source runtime developed by Nvidia for executing autonomous AI agents in isolated sandbox environments. It controls agents' access to files, processes, networks, external services, and credentials through kernel-level isolation and declarative security policies.[1][2]
OpenShell was publicly previewed for Ubuntu at Computex in June 2026 as part of a collaboration between Nvidia and Canonical.[2] Version 0.1.0 expanded the runtime with policy verification, credential-protected service access, support for multi-tenant deployments, and CPU and GPU execution.[3]
Architecture
OpenShell consists primarily of a gateway, supervisor, and sandbox. The gateway manages sandboxes and their policies, while the supervisor controls communication between an agent and external services.[4] Agent workloads execute inside sandboxes with operating-system controls restricting filesystem and process access.[3]
Policies are defined declaratively and can restrict network destinations and individual operations on services. OpenShell can also keep credentials outside the agent workload and provide them only to authorized requests. Its policy prover uses formal methods to analyze whether proposed permissions exceed defined security boundaries.[3]
OpenShell supports local and remote execution using Docker, Podman, MicroVMs, and Kubernetes.[5][6]
See also
- Artificial intelligence controversies
- Computer security
- List of artificial intelligence institutions
- List of cyberattacks
References
- ↑ "Overview of NVIDIA OpenShell". NVIDIA OpenShell Developer Guide. Nvidia. Retrieved September 28, 2026.
- ↑ 2.0 2.1 "Securing AI agent workflows on Ubuntu with the new NVIDIA OpenShell snap". Canonical. June 1, 2026. Retrieved September 28, 2026.
- ↑ 3.0 3.1 3.2 Watson, Alex; Golshan, Ali (September 28, 2026). "Add Runtime Controls to AI Agents with NVIDIA OpenShell". NVIDIA Technical Blog. Nvidia. Retrieved September 28, 2026.
- ↑ "Amid calls for 'pacing,' a new Nvidia safety tool to stop AI agents from going rogue". The Indian Express. 2026-09-29. Retrieved 2026-09-30.
- ↑ "Support Matrix". NVIDIA OpenShell Developer Guide. Nvidia. Retrieved September 28, 2026.
- ↑ "Nvidia says its new OpenShell platform can stop AI agents from going rogue - CBS News". www.cbsnews.com. 2026-09-28. Retrieved 2026-09-30.
External links
This article "Nvidia OpenShell" is from Wikipedia. The list of its authors can be seen in its historical and/or the page Edithistory:Nvidia OpenShell. Articles copied from Draft Namespace on Wikipedia could be seen on the Draft Namespace of Wikipedia and not main one.
