
CubeSandbox
Instant, Concurrent, Secure & Lightweight Sandbox for AI Agents.

Instant, Concurrent, Secure & Lightweight Sandbox for AI Agents.

Container escape proof-of-concept exploits for CVE-2026-80521 and CVE-2026-52910, with a disposable QEMU/Ubuntu VM harness for safe PoC execution.

Capability-based WASM runtime for executing untrusted AI-generated code with enforced CPU, memory, time, I/O, and filesystem limits. Provides…

Proof-of-concept demonstrating a Node.js permission model bypass (CVE-2026-21636) that allows network access via undici/fetch to local services,…

Rootless container runtime and sandbox that launches kernel-enforced OCI images in milliseconds with no daemon, featuring resource profiles, seccomp…

Secure and fast microVMs for serverless computing.

Sandboxes containers via a userspace application kernel that intercepts system calls, limits host kernel access, and integrates with…

CLI tool and library for generating a Software Bill of Materials from container images and filesystems

Qubes containerization on Windows

Securekit is a protocol-agnostic security kernel that enforces zero-trust, sandboxed execution for AI tool use. It sits between any LLM or agent…

A secure* runtime for autonomous AI agents. Policy from plain-English constitutions. (*https://ironcurtain.dev)

Kernel-level eBPF sandbox for securing LLM agent tool calls made through the Model Context Protocol (MCP)

In this project, we found a recent attack through the malicious container and implemented a security mechanism to stop it.

A lightweight, multi-layer Linux sandbox combining namespaces, pivot_root, seccomp-bpf, capability dropping, and an evidence-based verdict engine…

Run Firefox in a rootless Podman container with dropped capabilities, isolated networking, and ephemeral storage to contain sandbox escapes and…

Lightweight, secure Linux sandboxes for untrusted processes. Runs in the browser and on the server.

Run untrusted AI code safely, fast

A local sandbox for your AI agents