firecracker-sandbox

Upload a binary. Get a server.

MicroVM orchestrator for your VPS with built-in sleep/wake policies, backups, snapshots, HTTPS, custom image, logs and metrics

A lightweight microVM sandbox runtime for agents, automation, and untrusted workloads.

Virtual machine manager of Frappe Cloud

Bare-metal microVM execution engine in Rust. Sub-5ms snapshot restores, Zero-NIC network isolation, and 30 built-in language toolchains for AI code sandboxes & online judges.

koto is a microVM-based agent sandbox harness using Firecracker and that has the TUI and gRPC interface you always wanted

Cage lets you programmatically create isolated environments ("sandboxes"), run commands inside them, and tear them down — all through a simple REST API.

A Docker & Firecracker runtime for Apple Silicon Macs, powered by Lima

Secure, isolated sandbox execution on your own infrastructure

AgentBox SDK. AI sandboxes for company agents

CPU feature floor

Snapshot-optimised Linux guest for Firecracker microVMs. Measured post-restore working set: 1563 pages / 6.1 MiB. Open harnesses, WSL2-relative numbers.

Supply chain scanner for crates.io

Execute untrusted code securely in KVM microVMs with Firecracker, packaged as a single Rust binary for self-hosted judge and code execution.

Production security controls for Model Context Protocol (MCP) servers & AI agents: OAuth 2.1 audience verification, SHA-256 tool manifest pinning, gVisor/Firecracker sandboxing, and Kubernetes default-deny egress.

Run your GitHub Actions workflows locally in a fresh VM on every git push, and post the result to GitHub as a required check. Local CI for macOS, Linux and WSL2.

Threat-model-driven backup security blueprint: dual-control authorization, Firecracker-isolated repositories, PQC-ready signing, append-only restic storage. Design research — not a turnkey product.

Disposable, isolated microVMs with container ergonomics, for people and agents: Firecracker VMs from any OCI image, snapshots and fork, jailer and network isolation, web console, MCP server

Measure and tune the density of single-tenant Firecracker microVMs on Amazon EC2 metal (CPU pinning, NUMA-aware placement, labsweep harness) for RL rollouts and agentic code execution. Companion repo for the AWS Compute Blog post "Scaling RL rollouts with Firecracker on Amazon EC2 metal."

A beginner-friendly system design breakdown of safely running untrusted user code at scale in <50ms (LeetCode, Replit and AI Code Interpreter). Covers Linux cgroups v2, namespaces, seccomp filters and warm sandbox pools.