Armory
Source

Sandbox

Isolated compute an agent executes and deploys in

Listed
26
Ranked
53.8%
Top Score
99.9

Lists only rows made to run agent code in isolation: 26 of the 58 filed under Sandbox. The other 32 are in Browse.

Pick

The pick and its runners-up, in score order

  • daytona99.94 signals

    Runner-up · #1 on this shelf

    Sandboxes for AI-generated code that start in under 90 ms; its public repository stopped updating in June 2026

    71,846 stars · 5,651 forks · 9 mentions · passed install test
    No one-command install · Source
  • e2b-sandbox99.84 signals

    The pick · #2 on this shelf

    Firecracker microVM per run, ~150ms cold start, for untrusted code

    The one row above it is Daytona, whose public repository stopped updating in June 2026.

    13,635 stars · 1,015 forks · 10 mentions · passed install test
    No one-command install · Source
  • microsandbox99.12 signals

    Runner-up · #5 on this shelf

    Self-hosted microVM sandbox on your own machines, no per-sandbox vendor cost

    Above it, Daytona and E2B are mostly used as hosted services, container-use gives each coding agent a container on its own git branch, and Cua is built for computer-use agents; this one runs untrusted code in tiny virtual machines you host yourself.

    8,436 stars · 454 forks
    No one-command install · Source

Deploy access · E2B: E2B_API_KEY, one microVM per run. Every account

Top Ranked

Needs the armory CLI · not on npm yet, build it from cli/ in the repository

RankScoreComponentDescriptionEvidenceLast commitInstall
199.971daytonainfra · devopsSecure and elastic sandboxes for running AI-generated code. The public repository stopped updating in June 2026, when development moved to a private…71,846 stars · 5,651 forks · 9 mentions · passed install testNo one-command install · Source
299.886e2b-sandboxinfra · devopsUse as the default runtime when an agent must execute untrusted code or commands — Firecracker microVMs with ~150ms cold start give each run an…13,635 stars · 1,015 forks · 10 mentions · passed install testNo one-command install · Source
399.798container-useinfra · ai-agentsDevelopment environments for coding agents. Enable multiple agents to work safely and independently with your preferred stack.4,046 stars · 206 forks · passed install testNo one-command install · Source
499.470cua-computer-use-agentinfra · browsertrycua/cua open-source computer-use agent framework — Apple Silicon-native, runs lightweight macOS/Linux VMs with sub-second cold starts; provides a…22,092 stars · 1,519 forksNo one-command install · Source
599.100microsandboxinfra · devopsUse as the OSS self-hosted sandbox when you need to run agent code on your own infra — libkrun-based microVM isolation with no per-sandbox vendor…8,436 stars · 454 forksNo one-command install · Source
698.932claude-managed-agents-selfhostinfra · ai-agentsUse for enterprise hosted-control agents — the agent loop runs at the provider while execution happens on your own infrastructure — when you want a…3,876 stars · 836 forksNo one-command install · Source
798.698sysboxinfra · devopsNext-generation container runtime enabling Docker-in-Docker and VM-like isolation without privileged containers.3,847 stars · 228 forksNo one-command install · Source
898.124e2b-desktopinfra · browserE2B Desktop Sandbox — cloud virtual desktop (Ubuntu + VNC) with Python SDK for screenshot, mouse, keyboard, and process control; designed for AI…1,461 stars · 179 forksNo one-command install · Source
997.631surf-computer-useinfra · browserE2B Surf — a Stagehand-powered computer-use interface layer for E2B Firecracker sandboxes; connects the act/extract/observe primitives directly to…856 stars · 141 forksNo one-command install · Source
1097.091modal-labs-modal-clientinfra · devopsContributed by SentinelModal — serverless GPU/CPU containers for running agents, sandboxes and model inference from Python (this is the client SDK).518 stars · 132 forks · 11 mentionsNo one-command install · Source
1196.856runnoinfra · devopsExecute code in a secure WebAssembly sandbox supporting Python, JavaScript, C, C++, Ruby, and PHP772 stars · 48 forksNo one-command install · Source
1294.494babelcloud-gru-sandboxinfra · devopsSecure, self-hostable sandbox for running Python, TypeScript, and Bash code in isolated Docker containers with automatic resource management for local…180 stars · 50 forksNo one-command install · Source
1380.000e2binfra · devopsOpen-source secure cloud sandboxes (Firecracker microVMs) for running AI-generated code. ~150ms cold start.passed install testNo commit datelisted No one-command install · Source
1380.000modalinfra · back-endServerless cloud platform for running Python functions, containers, and AI workloads with zero infra management.passed install testNo commit datelisted No one-command install · Source

Score colour shows how many signals stand behind it, never how good it is: amber, three or more; dimmer amber, two; grey, one. The Evidence column names them.

Stars, forks and last commit are as GitHub reported them when Armory last read each repository: for most, or later. A repository may have changed since.

Leaderboard ranks all 14 scored rows. Formula shows the calculation.