使用 Matchlock 在一次性微虚拟机沙箱中运行代理，支持网络白名单和密钥注入
Launch risky agent work inside disposable microVMs when you need stronger isolation, sealed egress, and host-side secret injection instead of direct host access.
Prerequisites
Local shell, Matchlock CLI, virtualization support for the target host, and the agent image or command you want to run inside the microVM
Installation
Use the upstream install or setup path that matches your environment:
- brew tap jingkaihe/essentials
- brew install matchlock
- docker save myapp:latest | matchlock image import myapp:latest # Import from tarball
- pip install matchlock
Requirements and caveats from upstream:
- matchlock run --image python:3.12-alpine \
- --allow-host "api.openai.com" python agent.py
- --secret ANTHROPIC_API_KEY@api.anthropic.com python call_api.py
Basic usage or getting-started notes:
- AI agents need to run code, but giving them unrestricted access to your machine is a risk. Matchlock lets you hand an agent a full Linux environment that boots in under a second - isolated and disposable.
- ### System Requirements
- Linux with KVM support
- Source: https://github.com/1991513ccie-png/skills
- Extracted from upstream docs: https://raw.githubusercontent.com/jingkaihe/matchlock/HEAD/README.md
Documentation
- https://github.com/1991513ccie-png/skills
