thrashr888thrashr888Communityยท1 Agent Skills Included

agentkernel

Run coding agents in secure hardware-isolated microVM sandboxes

Runs coding agents and untrusted commands inside hardware-isolated microVM sandboxes with sub-125ms boot times. Eliminates the risk of generated code damaging your host system, leaking credentials, or escaping shared containers. Auto-detects project runtimes, enforces security profiles, and provides SSH access, snapshots, and parallel job execution for faster safe testing.
npx skills add thrashr888/agentkernel --all -g -y
Available:

Instructs your AI agent on how to track work with bd issues, follow mandatory session-completion and push rules, and use the sandbox build, test, and security-profile commands correctly.

All Skills in This Repository (1)

Pure Emerald Level Indicators

Frequently Asked Questions

FAQPage Schema
How to install agentkernel?โ–ผ

Run `npx skills add thrashr888/agentkernel --all -g -y` in your terminal to install all skills in this suite globally.

How to run untrusted code safely?โ–ผ

Use `agentkernel run <command>` to execute any command inside an isolated microVM sandbox with its own Linux kernel, so nothing can affect your host system.

How is agentkernel different from Docker containers?โ–ผ

Each sandbox runs in its own virtual machine with a dedicated kernel via KVM and Firecracker, instead of sharing the host kernel like Docker, giving much stronger isolation.

Does agentkernel work with Claude Code and other coding agents?โ–ผ

Yes. It supports Claude Code, Gemini CLI, Codex, and OpenCode through plugins, an MCP server, and per-agent compatibility modes with tuned network and permission rules.

Can I control sandbox network and file access?โ–ผ

Yes. Choose from permissive, moderate, or restrictive security profiles, or use flags like --no-network to block network access and keep the filesystem read-only.

Related Repositories in Software Engineering

View All in Software Engineeringโ†’