mcpbeat Sign in

LLM Sandbox MCP Server

local only

LLM Sandbox runs on your own machine — the client starts it, so there is no endpoint to ping. 122 913 installs a week from pypi. Last commit 26 Oct 2026.

Securely run LLM-generated code in isolated containers across 7 languages and 3 container backends.

Installs per day peak 22 990 · avg 14 330 · +21% w/w
a month agotoday
122 913
Installs / week
pypi · llm-sandbox
1 139
Stars
19 open issues
26 Oct 2026
Last commit
from GitHub
MIT
License
Python

What the code does

We read the source, 6 d ago · rules 995398eba69f

Capabilities

What this server is able to do. For an MCP server this is often the job itself — a terminal server runs commands because that is what it is for. Listed so you know what you are plugging in, not as an accusation.

Is this your server and something here is wrong? Tell us — corrections are free and do not require a plan.

This code can reach further than it looks

We found places where it runs commands, builds paths or queries from values it is given. None of that is a flaw by itself — it becomes one when the code changes, and code changes quietly between releases. We re-read it on every one.

Three servers free · no card

Connect this server

This server runs on your own machine — install it with the package manager and the client starts it for you. Package name taken from the official registry entry.

run in your terminal
claude mcp add llm-sandbox -- uvx llm-sandbox
~/Library/Application Support/Claude/claude_desktop_config.json
{
  "mcpServers": {
    "llm-sandbox": {
      "command": "uvx",
      "args": [
        "llm-sandbox"
      ]
    }
  }
}
~/.codex/config.toml
[mcp_servers.llm-sandbox]
command = "uvx"
args = ["llm-sandbox"]
.cursor/mcp.json
{
  "mcpServers": {
    "llm-sandbox": {
      "command": "uvx",
      "args": [
        "llm-sandbox"
      ]
    }
  }
}
.vscode/mcp.json
{
  "servers": {
    "llm-sandbox": {
      "type": "stdio",
      "command": "uvx",
      "args": [
        "llm-sandbox"
      ]
    }
  }
}

Optional settings from the registry entry, left out of the block above: BACKEND (Container backend to use. Must match the installed extra: mcp-docker, mcp-podman, or mcp-k8s), DOCKER_HOST (Docker or Podman socket URL, e.g. unix:///var/run/docker.sock), KUBECONFIG (Path to kubeconfig file when BACKEND=kubernetes), NAMESPACE (Kubernetes namespace used for sandbox pods when BACKEND=kubernetes), COMMIT_CONTAINER (Commit the container after a run so installed libraries persist between sessions), KEEP_TEMPLATE (Keep the base image after the session ends to avoid re-pulling it on the next run), SANDBOX_NETWORK_MODE (Network mode for the sandbox container. Set to 'none' for hardened isolation. Docker and Podman backends only), SANDBOX_READ_ONLY (Mount the sandbox root filesystem read-only. Recommended: true. Docker and Podman backends only), SANDBOX_CAP_DROP (Comma-separated Linux capabilities to drop. Recommended: ALL. Docker and Podman backends only), SANDBOX_SECURITY_OPT (Comma-separated container security options, e.g. no-new-privileges. Docker and Podman backends only), SANDBOX_MEMORY (Memory limit for the sandbox container, e.g. 4g. Docker and Podman backends only), SANDBOX_CPUS (Fractional CPU allocation for the sandbox container, e.g. 1.5. Docker and Podman backends only).

Alternatives to LLM Sandbox

same job, measured the same way
Cinch
by yusufkadry

Run AI-generated code safely. gVisor-isolated Python/JS sandboxes for any MCP client. Pay per run.

33 installs/wk local only
Cdbx
by cdbx

Build Apps and run code in 30 languages — sandboxed, with persistent sessions for agent loops.

answering
H
Hopx MCP
by hopx-ai

Execute code in isolated cloud containers. Supports Python, JavaScript, Bash, and Go.

local only
Agent Lsp
by blackwell-systems

Orchestrates language servers into 65 code-intelligence tools across 32 languages, token-optimized.

1 474 installs/wk local only
MetaEngine
by metaengine

Code generation from JSON recipes, linked type graphs, and API schemas across 11 languages.

67 installs/wk local only
E2b
by mcparmory

Create and manage secure sandboxes for running AI agent code

207 installs/wk local only
SandboxAPI
by michielderuiter

Execute code in 8 languages (Python, JS, TS, Go, Java, C++, C, Bash) in gVisor sandboxes.

answering
Mathematical Code Execution
by saifs

Run sandboxed Python for calculations and PDFs; no other languages, installs or network.

answering

LLM Sandbox — questions

Answers built from our own checks of this server.

Why is there no uptime for LLM Sandbox?
LLM Sandbox runs on your own machine over stdio — there is no network address to reach, so uptime cannot be measured for it by anyone. What can be measured is adoption: the pypi package llm-sandbox was installed 122 913 times last week.
How do I connect LLM Sandbox?
Copy the ready config from this page. We generate it for Claude Code, Claude Desktop, Codex, Cursor and VS Code, each in the form that client actually reads. It runs locally, so the command pulls llm-sandbox straight from pypi; nothing to host, nothing to sign up for.
How many people use LLM Sandbox?
The pypi package llm-sandbox was installed 122 913 times in the last week. Week over week that is +21%. We show installs rather than GitHub stars on purpose: a star is a bookmark, an install is someone actually running it.
Is LLM Sandbox open source?
Yes — it is published under the MIT licence, written in Python, 1 139 stars on GitHub and 19 open issues. The source link is on this page, so you can read exactly what it does with your data before you connect it.