npm package discovery and stats viewer.

Discover Tips

  • General search

    [free text search, go nuts!]

  • Package details

    pkg:[package-name]

  • User packages

    @[username]

Sponsor

Optimize Toolset

I’ve always been into building performant and accessible sites, but lately I’ve been taking it extremely seriously. So much so that I’ve been building a tool to help me optimize and monitor the sites that I build to make sure that I’m making an attempt to offer the best experience to those who visit them. If you’re into performant, accessible and SEO friendly sites, you might like it too! You can check it out at Optimize Toolset.

About

Hi, 👋, I’m Ryan Hefner  and I built this site for me, and you! The goal of this site was to provide an easy way for me to check the stats on my npm packages, both for prioritizing issues and updates, and to give me a little kick in the pants to keep up on stuff.

As I was building it, I realized that I was actually using the tool to build the tool, and figured I might as well put this out there and hopefully others will find it to be a fast and useful way to search and browse npm packages as I have.

If you’re interested in other things I’m working on, follow me on Twitter or check out the open source projects I’ve been publishing on GitHub.

I am also working on a Twitter bot for this site to tweet the most popular, newest, random packages from npm. Please follow that account now and it will start sending out packages soon–ish.

Open Software & Tools

This site wouldn’t be possible without the immense generosity and tireless efforts from the people who make contributions to the world and share their work via open source initiatives. Thank you 🙏

© 2026 – Pkg Stats / Ryan Hefner

@husk-ai/mcp

v0.1.4

Published

Give any MCP client a Linux computer.

Readme

@husk-ai/mcp

Give any MCP client a Linux computer.

claude mcp add husk -- npx -y @husk-ai/mcp

That is the whole setup. No account, no config file, no API key. Claude Code now has a shell, a filesystem and ports on a real Linux machine, and the filesystem persists for the whole conversation.

Works the same in Cursor, Zed, or anything else that speaks MCP:

{
  "mcpServers": {
    "husk": { "command": "npx", "args": ["-y", "@husk-ai/mcp"] }
  }
}

Tools

| tool | what it does | | --- | --- | | shell | run a command; /work persists across calls | | read_file | read a text file, with optional line range | | write_file | create or overwrite, creating parent directories | | edit_file | exact-string replace; fails loudly if the match is missing or ambiguous | | list_dir | list a directory | | expose_port | publish a port and get a reachable URL | | computer_info | OS, kernel, CPU, memory, disk, network, installed runtimes |

computer_info exists so a model spends one turn learning the machine instead of four probing it with uname, nproc, df and which python3.

Options

--session <key>     reuse a named machine (default: unique to this server process)
--provider <name>   docker | podman | local | ssh | fly
--flavor <name>     base | python | node | full
--network <mode>    none | egress | full
--memory <mb>       memory ceiling
--cpus <n>          cpu ceiling
--keep              leave the machine running after the client disconnects

Two projects, two machines:

claude mcp add husk-api  -- npx -y @husk-ai/mcp --session api  --flavor python
claude mcp add husk-web  -- npx -y @husk-ai/mcp --session web  --flavor node

What kind of machine you get

Husk picks the best provider available and tells the model which one it got, in the first tool result:

[husk] docker container cmp_8v9z, isolated from the host. /work persists for this session.

or, with no Docker running:

[husk] local computer cmp_8v9z on real Linux via wsl:Ubuntu. This is a guarded working
directory, NOT a sandbox: file tools are confined to /work and destructive commands are
refused, but shell commands can still reach anything your user can on the host --
filesystem, kernel and network. Keep secrets and untrusted input out of it.

A model that believes it is sandboxed when it is not makes worse decisions than one that knows, so the note is not optional and it is not softened.

Cost of installing it

Nothing is created until the first tool call. Adding the server starts a Node process that registers seven tool schemas and waits — no container, no directory, no network.

Verify your install

npx -y @husk-ai/mcp --help      # options, to stderr
node dist/smoke.js           # drives every tool against a real machine

The smoke test creates a computer, exercises each tool, checks that a destructive command is refused and that a path outside /work is rejected, then destroys the machine.

Notes

  • stdout is the JSON-RPC stream. All logging goes to stderr; a stray console.log would corrupt the protocol and silently kill the client.
  • Closing stdin shuts the server down and destroys the machine, so a crashed client does not leave a container running.
  • --keep opts out of that, for a machine you want to reattach to.

Reconnects and health checks

A stopped computer is started again before use, preserving its workspace. The MCP server checks a cached computer before each tool call. An unavailable engine or a failed health probe is reported as an error; a connection alone does not mean the computer is ready. Run husk doctor under the same account as the MCP client to check provider access. On Windows, start Docker Desktop and check docker info.

Each server process gets its own session key by default so separate clients do not stop or reuse each other’s computer. Use --session <key> --keep (or HUSK_SESSION with --keep) to reconnect to a named workspace. Clients that explicitly share a key must use --keep to avoid destroying it on disconnect. A client that shares one MCP process across chats also shares that computer.

The optional viewer is separate: viewer not running only means husk serve is not available. It does not describe the container’s health.