LLM Sandbox MCP Server

Securely Execute LLM-Generated Code with Ease LLM Sandbox is a lightweight and portable sandbox environment designed to run Large Language Model (LLM) generated code in a safe and isolated mode. It provides a secure execution environment for AI-generated code while offering flexibility in container backends and comprehensive language support, simplifying the process of running code generated by LLMs. Its 3 documented tools cluster into read & search, run & execute.

People connecting coding agent tools to Claude, Cursor, VS Code, or another MCP client. The project is written in Python.

VERIFIED ACTIVE

LAST COMMIT 2026-08-23 · ★ 1,111 · #3 OF 194 MAINTAINED CODING AGENT · VERIFIED 2026-08-25

MIT · Python servers · how we verify → /methodology

01 · Install LLM Sandbox

before you install - you'll need

Set DOCKER_HOST, SANDBOX_NETWORK_MODE, SANDBOX_READ_ONLY, SANDBOX_CAP_DROP, SANDBOX_SECURITY_OPT, SANDBOX_MEMORY, SANDBOX_CPU_COUNT before connecting.

Claude Code

claude mcp add vndee-llm-sandbox -- uvx llm-sandbox

Claude Desktop / Cursor / VS Code - add to config

{
  "mcpServers": {
    "vndee-llm-sandbox": {
      "command": "uvx",
      "args": [
        "llm-sandbox"
      ]
    }
  }
}

Same JSON for Cursor. For VS Code, rename the top-level key from `mcpServers` to `servers`.

Using another client? Same JSON, different key

Claude Desktop · mcpServers

Cursor · mcpServers

VS Code · servers

Windsurf · mcpServers

Zed · context_servers

Cline · mcpServers

Roo Code · mcpServers

Continue · mcpServers

LibreChat · mcpServers

Gemini CLI · mcpServers

Codex CLI · mcp_servers

Full setup guides: every client.

02 · Evidence

Security posture

What to check before giving this server access to your agent - from the registry, GitHub, and our own probes. We don't score safety; we show what's verifiable.

runs as local process (stdio) - runs on your machine with your user's permissions

license MIT - declared in the repository

pypi package llm-sandbox - check the name against the project README before installing (PyPI has no namespace ownership)

registry namespace io.github.vndee is GitHub-verified and matches the repo owner

03 · What LLM Sandbox can do

Prose above is summarized from the project's README and registry record - no invented capabilities.

What you can build

With this server connected, an agent can execute code in a secure sandbox with automatic visualization capture, and get detailed information about a specific language.

The 3 tools it gives your agent

Extracted from the project's README - what LLM Sandbox lets an agent do.

execute_code: - Execute code in a secure sandbox with automatic visualization capture
get_supported_languages: - Get the list of supported programming languages
get_language_details: - Get detailed information about a specific language

Latest releases

0.3.44 · 2026-08-03

Add JOSS required sections to the paper · Open a PR from the citation job instead of pushing to main · Point the registry entry at 0.3.43 · Add agent SDK integrations and repair the broken ones · Smoke-check the agent…

0.3.43 · 2026-08-03

Fixes a broken MCP install · mcp 2.0.0 removed mcp.server.fastmcp, which the MCP server imports at module scope. The extras allowed it through an unbounded mcp>=1.10.0, so any fresh install since mcp 2.0.0 shipped…

0.3.42 · 2026-08-02

Add MCP registry manifest, JOSS paper, and narrative docs page

04 · Who maintains LLM Sandbox

LLM Sandbox is maintained by vndee. It's the only MCP server we track from this author; the repo dates to Jun 2024.

05 · Facts

repository: github.com/vndee/llm-sandbox
website: https://vndee.github.io/llm-sandbox/
category: coding agent - ranked #3 of 194 actively-maintained coding agent servers as of 2026-08-25.
release cadence: 5 releases in the last 90 days (latest 2026-08-03)
registry: io.github.vndee/llm-sandbox (active, first published 2026-08-02 · 2 versions)
packages: pypi:llm-sandbox

06 · LLM Sandbox FAQ

Is LLM Sandbox still maintained?

Yes - as of 2026-08-25, its last commit was 2026-08-23 and it shipped 5 releases in the last 90 days. We re-verify nightly.

What can LLM Sandbox do?

With this server connected, an agent can execute code in a secure sandbox with automatic visualization capture, and get detailed information about a specific language.

How do I install LLM Sandbox?

Run `uvx llm-sandbox`. The README documents 7 environment variables (DOCKER_HOST, SANDBOX_NETWORK_MODE, SANDBOX_READ_ONLY…) to set first. Set DOCKER_HOST, SANDBOX_NETWORK_MODE, SANDBOX_READ_ONLY, SANDBOX_CAP_DROP, SANDBOX_SECURITY_OPT, SANDBOX_MEMORY, SANDBOX_CPU_COUNT before connecting. You can also paste the ready-made client config above.

Does LLM Sandbox run locally?

Yes - it's a stdio server: it runs on your machine (via uvx) with your user's permissions. Your data stays local unless the server itself calls external APIs.

07 · Alternatives to LLM Sandbox

Alternatives to LLM Sandbox

Maintained coding agent servers if LLM Sandbox isn't the fit.

Pairs well with

Servers that cover what LLM Sandbox doesn't - only shown when the pairing reason fits the companion.

head to head · Serena vs LLM Sandbox · LLM Sandbox vs Desktop Commander · Ouroboros vs LLM Sandbox

More coding agent MCP servers · Deja Vu · VULK · MetaMask Embedded Wallets (Web3Auth) MCP · Exerciseapi · Alive Analysis

More Python MCP servers · Yggdrasil · Midas · Outline MCP Server · Vouch MCP · Vouch · see all