Skip to content
English

Switch theme

Discover▾

Not sure where to start? This week’s hottest software, models and stacks in three minutes.

AI Apps▾

AI software people use: from assistants like ChatGPT to tools for coding, video and workflows, picked by purpose.

For Agents▾

Software for AI: tools that let your agent (Claude Code, Codex, Cursor…) browse, read docs, run code and remember.

Models▾

Choose models, estimate cost, self-host: prices, rankings, benchmarks, free tiers and compute.

Solutions▾

A goal or an error message? Ready-made stacks, steps and fixes by scenario.

Learn▾

A systematic course from zero to the frontier, with an algorithm library, must-read papers and tutorials.

Frontier▾

Daily AI news, weekly highlights, the latest papers and how AI is changing every discipline.

Tools▾

Small tools that run right in your browser: your data stays on your device.

// Run code in a sandbox

NVIDIA OpenShell

NVIDIA’s open-source safe runtime for autonomous agents (latest stable v0.1.5, 2026-10-20): each agent runs in an isolated sandbox with kernel-level control of file access, system calls and every network connection; before a policy change is applied, formal verification flags risky new access (e.g. reaching a new host with credentials) for human review. Credentials only work at approved endpoints; ships a gateway, Kubernetes Helm deployment, Python/TypeScript/Go/Rust SDKs and agent skills. Needs Docker, Podman or host virtualization

Open source Maintained For agents

Official site ↗ GitHub ↗

OFFICIAL DOWNLOADDirect official link

v0.1.5 · 2026-10-20

Download Mac · Apple silicon9.1 MB · openshell-aarch64-apple-darwin.tar.gz Download Linux · x6475.9 MB · openshell_0.1.5-1_amd64.deb

Every button links straight to the official file (github.com). AIMap does not host or modify installers. Checked 2026-10-27. All files, checksums & install steps →

IN ONE PARAGRAPH

NVIDIA OpenShell is NVIDIA’s open-source safe runtime for autonomous agents (latest stable v0.1.5, 2026-10-20): each agent runs in an isolated sandbox with kernel-level control of file access, system calls and every network connection; before a policy change is applied, formal verification flags risky new access (e.g. reaching a new host with credentials) for human review. Credentials only work at approved endpoints; ships a gateway, Kubernetes Helm deployment, Python/TypeScript/Go/Rust SDKs and agent skills. Needs Docker, Podman or host virtualization.

Good for: Developers who want their agent (Claude Code, Codex, Cursor…) to run code in a sandbox.

Not for: people who don’t want to install and run tools themselves.

01How to start

  1. 1Read the official README for requirements and permissions.
  2. 2Install it as the official guide describes.
  3. 3Tell your agent the command exists (e.g. in AGENTS.md / CLAUDE.md) so it knows when to call it.

02Alternatives & comparison

Compare side by side →
firecracker Secure and fast microVMs for serverless computing. ★ 37k gvisor Application Kernel for Containers ★ 20k pyodide Pyodide is a Python distribution for the browser and Node.js based on WebAssembly ★ 15k E2B Open-source, secure environment with real-world tools for enterprise-grade agents. ★ 15k kubernetes Production-Grade Container Scheduling and Management ★ 128k

03Related

Versions, why it is listed, for AI to read

For AI agents

Data updated: AI software & hot ranks · Software profiles & needs · Official software downloads(UTC+8)