Allows AI agents to manage a Linux server through shell sessions, including background execution and sudo separation.
Click on "Deploy Server".
Wait a few minutes for the server to deploy. Once ready, it will show a "Started" state.
In the chat, type @ followed by the MCP server name and your instructions, e.g., "@Linux Manager MCP Check system load and disk space, then show me the top memory processes"
That's it! The server will respond to your query, and you can continue using it as needed.
Here is a step-by-step guide with screenshots.
This is a for-fun project. Do not use it in production, and do not use it on any machine you care about: you are on your own if you do. No security guarantees are made. Friend link: Linux.do / A collection of AI-wiped-my-server incidents seen on the forum (login required): https://linux.do/t/topic/1585027
Give your LLM access to your Linux server and let it drive. Unlike OpenClaw, this is an lite MCP plugin: it needs no API key, and for now it has no memory either.
Related MCP server: RemoteShell MCP
English: Demo
中文： 用法示例
Multiple shell windows, with background execution.
When you ask an LLM to install a dependency, a conventional MCP will just keep running until the command finishes or the timeout kills it.
This project lets the LLM open several shell windows, set its own foreground wait time and hard kill time, and pull the output so far at any moment to see how things are going.
Every shell window has a notes area.
So the LLM can remind itself why a window was opened and what has been done in it.
Sudo separation, no sudo privileges by default.
Shell sessions handed to the LLM are non-sudo; it has to explicitly ask for sudo to get it.
Sudo can be turned off entirely in the program.
Ctrl+C to interrupt a command (instead of killing the whole shell), finer-grained controls, a logging system, several people controlling the same machine.
This server cannot be deployed
Operate Linux, macOS and Windows from your LLM. Every action runs through an auditable allowlist.
Let AI operate servers without SSH. Choose actions, approve risky changes, and audit every step.
A persistent Linux computer for your AI: what it installs and writes survives the session.
Shared control plane for AI coding agents — tasks, memory, decisions, file locks. 12 tools.
