Token导航 LogoToken导航TokenDH.com
开发执行命令github未标认证来源可访问许可证需确认审计异常

microsandboxmicrosandbox 命令行

Agent Skill

microsandbox 用于处理 GitHub 仓库、Issue、Pull Request 和代码协作信息,适合在 Codex、Claude、Cursor、Gemini CLI 中需要围绕仓库状态、代码变更或协作事项进行整理时使用。可结合来源仓库、安装命令和原始 README 继续核验具体用法。安装前建议确认权限范围、维护状态,以及是否会触发联网、命令执行或文件读写。

总安装

706

周安装

30

GitHub Stars

5

下载量

247
CodexClaudeCursorGemini CLI

安装说明

本站只整理中文说明和来源信息,不托管安装包,也不代用户安装。

GitHub

来源数

2

许可证

unknown

最后核验

2026-05-01

来源状态

来源可访问

安装方式

通过对话安装

复制提示词发给支持本地命令或 Skills 的 AI 助手,先确认命令和权限,再让它执行。

请帮我安装这个 Agent Skill:microsandbox(microsandbox 命令行)
来源仓库:https://github.com/superradcompany/skills
仓库路径:skills/microsandbox
安装命令:
npx skills add https://github.com/superradcompany/skills --skill microsandbox
安装前请先检查当前环境是否支持对应 CLI,并向我确认将要执行的命令、安装目录、联网范围和文件读写权限;确认后再执行。

命令行安装

复制命令到本机终端执行。该命令会通过 npx skills 从第三方来源获取 Skill;本站只展示命令,不托管安装包,也不自动执行。

skills.shnpx skills
npx skills add https://github.com/superradcompany/skills --skill microsandbox

简介

用于处理 GitHub 仓库、Issue、Pull Request 和代码协作信息。

  • 适合围绕仓库状态、代码变更进行整理。
  • 可结合来源仓库和原始 README 核验具体用法。
  • 安装前建议确认权限范围和维护状态。microsandbox 属于开发类 Skill,可作为该场景下的辅助能力补充。
  • 注意是否会触发联网、命令执行或文件读写。

SKILL.md

microsandbox

microsandbox creates hardware-isolated microVMs that boot in under 100ms. Each sandbox is a real VM with its own Linux kernel — not a container.

Setup

Check if microsandbox is installed:

msb --version

If not installed, run the setup script:

bash scripts/setup.sh

This installs msb to ~/.microsandbox/bin/ and libkrunfw to ~/.microsandbox/lib/.

Quick reference

Run a one-off command in a sandbox

msb run <image> [options] -- <command>

Examples:

msb run python:3.12 -- python -c "print('hello from sandbox')"
msb run -m 1G node:22 -- node -e "console.log(process.version)"
msb run alpine:latest -- sh -c "uname -a && cat /etc/os-release"

Create a persistent sandbox

msb create --name <name> [options] <image>
msb exec <name> -- <command>
msb shell <name>
msb stop <name>
msb start <name>                  # Resume a stopped sandbox
msb rm <name>

Example workflow:

# Create a Python development sandbox
msb create --name dev -m 1G -c 2 python:3.12

# Install packages
msb exec dev -- pip install requests numpy

# Run code
msb exec dev -- python -c "import requests; print(requests.get('https://httpbin.org/ip').json())"

# Interactive shell
msb shell dev

# Stop and resume later
msb stop dev
msb start dev

# Clean up
msb stop dev
msb rm dev

Common options

FlagDescriptionExample
-n, --nameName the sandbox--name my-sandbox
-m, --memoryMemory allocation-m 512M, -m 1G
-c, --cpusNumber of vCPUs-c 2
-v, --volumeMount volume-v /host/path:/guest/path
-p, --portPublish port-p 8080:80, -p 5353:5353/udp
-e, --envSet env variable-e API_KEY=xxx
-w, --workdirWorking directory-w /app
-d, --detachRun in background (run only)-d
-u, --userRun as user-u nobody
-H, --hostnameSet guest hostname-H myhost
--shellDefault shell program--shell /bin/bash
--replaceReplace existing sandbox--replace
--entrypointOverride entrypoint--entrypoint /bin/sh
--pullPull policy--pull always
--max-durationAuto-stop timeout--max-duration 5m
--idle-timeoutIdle auto-stop--idle-timeout 30s
--tmpfsMount tmpfs--tmpfs /tmp:100M
--scriptInject script--script setup:./setup.sh

Manage sandboxes

msb ls                    # List all sandboxes
msb ls --running          # Running only
msb ps                    # Show running sandboxes with status
msb ps -a                 # All sandboxes including stopped
msb inspect <name>        # Detailed sandbox info
msb metrics <name>        # Live CPU/memory/IO stats
msb stop <name>           # Graceful shutdown
msb stop --force <name>   # Force kill
msb stop -t 10 <name>    # Wait 10s then force kill
msb rm <name>             # Remove stopped sandbox
msb rm --force <name>     # Stop and remove in one step

Manage images

msb pull <image>          # Pre-cache an OCI image
msb images                # List cached images (alias: msb image ls)
msb image inspect <img>   # Image metadata
msb rmi <image>           # Remove cached image (alias: msb image rm)

Manage volumes

msb volume create <name>          # Create named volume
msb volume create <name> --size 5G  # With quota
msb volume ls                     # List volumes
msb volume inspect <name>         # Volume details
msb volume rm <name>              # Remove volume

Volume mounts

# Bind mount host directory
msb run -v ./project:/app python:3.12 -- python /app/script.py

# Named volume (persistent across sandboxes)
msb volume create mydata
msb run -v mydata:/data alpine -- sh -c "echo 'test' > /data/file.txt"
msb run -v mydata:/data alpine -- cat /data/file.txt

Networking and security

# No network access
msb run --no-network python:3.12 -- python script.py

# Block specific domains
msb run --dns-block-domain "ads.example.com" python:3.12

# Inject secrets (placeholder substitution — real value never enters VM)
msb run --secret "OPENAI_API_KEY=sk-xxx@api.openai.com" python:3.12

# TLS interception for secret injection
msb run --tls-intercept --secret "API_KEY=xxx@api.example.com" python:3.12

# Limit connections
msb run --max-connections 10 python:3.12

Registry authentication

msb registry login ghcr.io --username octocat
msb registry logout ghcr.io
msb registry ls

Install sandbox as command

msb install python:3.12          # Install as 'python' command
msb install --name py python:3.12  # Custom name
msb install --list               # Show installed commands
msb uninstall py                 # Remove

Key behaviors

  • Sandboxes are real microVMs with hardware-level isolation (hypervisor boundary)
  • Boot time is under 100ms
  • Default network policy is public-only (blocks private ranges, metadata endpoints)
  • Sandboxes from msb run without --name are ephemeral (destroyed after exit)
  • Sandboxes from msb create or msb run --name are persistent (survive until msb rm)
  • msb create always runs in background; use msb run -d for detached one-off runs
  • Secrets use placeholder substitution — real credentials never enter the VM
  • Use --replace to recreate an existing sandbox with new settings

Troubleshooting

If msb is not found after installation:

source ~/.bashrc   # or ~/.zshrc

Check installation:

ls ~/.microsandbox/bin/msb
ls ~/.microsandbox/lib/libkrunfw*

For full CLI reference, see references/cli-reference.md. For SDK usage, see references/sdk-typescript.md and references/sdk-rust.md.

适合场景

01

用户想查找某类 Agent Skill 时

02

需要根据任务场景推荐可安装能力包时

03

需要对比不同来源的安装命令和来源信息时

能力概览

能力 1

按任务关键词查找相关 Skills

能力 2

展示可复制的安装命令

能力 3

保留来源站点、仓库和原始说明,方便继续核验

能力 4

展示第三方安全扫描或审计结果

安装后应在对应宿主中按原始 README 的触发条件使用;具体调用方式请以来源页面和 README 为准。

平台分布

Codex

34.38%
按下载量换算85

Claude

31.13%
按下载量换算77

Cursor

17.97%
按下载量换算44

Gemini CLI

10.19%
按下载量换算25

安全审计

Gen Agent Trust Hub

未通过

Socket

可疑

Snyk

未通过

权限和风险

执行命令

安装流程涉及命令执行,可能通过 npx skills add https://github.com/superradcompany/skills --skill microsandbox 联网下载 Skill 或依赖。用户安装前应确认命令来源、仓库内容和执行环境。

安装前确认

本站仅展示第三方公开信息,不托管安装包,不提供自动安装或运行环境。安装前应自行审查源码、依赖和命令行为。来源安全扫描存在 warning/failed 结果,不能写成本站确认安全。当前只有一个来源,正式发布前建议补源仓库或其他目录站核验。

来源信息

继续浏览同类 Skills