Skip to content
MCP server

WBX Sandbox

Firecracker microVM sandboxes your assistant can use to run code, jobs and batches

Install
Pricing
Freemium
Platforms
API, Web
Runs
Remotely
Version
0.14.21
Listing updated
2 days ago

About WBX Sandbox

WBX Sandbox is a freemium MCP server for cloud and infrastructure. It runs as a hosted remote server, so there is nothing to install locally.

WBX Sandbox is hosted in the EU and reached through a streamable-HTTP MCP endpoint with OAuth sign-in, so ChatGPT, Claude or Claude Code can start a fresh Linux machine for a task and delete it when the work is done. The sandbox has internet access by default, which can be limited or turned off, and full root so an assistant can install tools with pip, npm or apt.

The site stresses keeping heavy work out of the chat. Test runs and long logs are condensed into a short digest (recognised test results plus errors with context), the assistant can wait on the server for a job instead of polling, and large files never pass through the conversation. Each background job or batch carries a price cap that Budget Guard checks before it starts, and every run produces a receipt with runtime, isolation, resource use, file hashes and actual cost.

Developers get the same platform through a REST API and Python SDK, with batches of up to 1,000 jobs with dependencies, WebAssembly alongside microVMs, and a LangChain deep agents integration.

Install

Run in a terminal

claude mcp add --transport http sandbox https://sandbox.wbx.cz/mcp

Add to Settings → Connectors → Add custom connector

https://sandbox.wbx.cz/mcp
Install in CursorOpens Cursor, which asks you to confirm. Or add it by hand:

Add to ~/.cursor/mcp.json

{
  "mcpServers": {
    "sandbox": {
      "url": "https://sandbox.wbx.cz/mcp"
    }
  }
}
Install in VS CodeOpens VS Code, which asks you to confirm. Or add it by hand:

Add to .vscode/mcp.json

{
  "servers": {
    "sandbox": {
      "type": "http",
      "url": "https://sandbox.wbx.cz/mcp"
    }
  }
}

Add to ~/.codeium/windsurf/mcp_config.json

{
  "mcpServers": {
    "sandbox": {
      "serverUrl": "https://sandbox.wbx.cz/mcp"
    }
  }
}

Add to your client's MCP settings (most use this shape)

{
  "mcpServers": {
    "sandbox": {
      "type": "http",
      "url": "https://sandbox.wbx.cz/mcp"
    }
  }
}

Tools this server provides (11)

Taken from the project documentation; the running server may differ.

create_sandbox
Start a sandbox.
run_command
Run a shell command in a sandbox.
run_code
Run code in a sandbox.
write_file
Write a file into a sandbox.
run_job
Submit a background job.
run_batch
Submit a batch of jobs.
wait_jobs
Wait on the server for jobs to finish.
get_job_logs
Read job logs, with a summary mode.
get_budget
Check the remaining budget.
get_compute_classes
List compute classes.
download_job_artifact
Download a job's output file.

What it does

Disposable microVMs
Each task gets its own Linux machine that is deleted afterwards.
Compact job logs
Failed runs return a summary of test results and errors rather than thousands of log lines.
Budget Guard
Worst-case cost is reserved before a job or batch starts and refused if it would break your limit.
Batches with dependencies
Up to 1,000 jobs per batch, where one job's result files feed the next.
Execution receipts
Per-job record of runtime, isolation, network policy, file hashes and cost, kept after the sandbox is gone.
Two runtimes
Full Linux microVMs and lightweight WebAssembly share one API.

Good for

  • Have an assistant run Python to analyse a spreadsheet and draw a chart
  • Clone a repository, run its tests and report only failures
  • Process hundreds of files overnight within a spending cap
  • Run an untrusted script with no network to see what it does
  • Give a LangChain deep agent an isolated place to run commands

Pricing

Freemium

Pay as you go from $0.038 per vCPU-hour and $0.013 per GB-hour; welcome credit; fixed monthly prices for Solo, Pro and Team not published

Prices change. Check sandbox.wbx.cz before you buy.

WBX Sandbox FAQ

How do I connect it to Claude Code?

Run `claude mcp add --transport http wbxsandbox https://sandbox.wbx.cz/mcp`, then sign in through /mcp.

Do I need a card to try it?

The site says no card is required to start, and new accounts get welcome credit.

Are there fixed plans?

Yes: pay as you go, plus Solo, Pro and Team plans with different concurrency, RAM and time limits. Monthly prices for the fixed plans are not shown on the page.

Where does it run?

On servers in the EU. Pay-as-you-go usage is billed per second, at $0.038 per vCPU-hour and $0.013 per GB-hour.

Do budget limits cover everything?

Budget Guard covers background jobs and batches. Sandboxes you start directly are billed from your credit and stop starting when it runs out.

Is WBX Sandbox free?

WBX Sandbox has a free plan, with paid plans for more usage or features.

How do I add WBX Sandbox to Claude Code?

Run this command in your terminal: claude mcp add --transport http sandbox https://sandbox.wbx.cz/mcp

What are the best alternatives to WBX Sandbox?

The closest alternatives to WBX Sandbox are REA, Codebase Memory, GitHub, Serena MCP: the IDE for your agent and XERJ.

Alternatives to WBX Sandbox

  1. REA

    MCP server

    CLI and MCP server that lets coding agents reverse engineer apps and binaries

    Runs locally
    • Open source
    • 464K GitHub stars
    • Developer tools
  2. Codebase Memory

    MCP server

    Local MCP server that builds a code knowledge graph for AI coding agents

    Runs locally
    • Open source
    • 48.7K GitHub stars
    • Knowledge and memory
  3. GitHub

    MCP server

    Official MCP server connecting AI assistants to GitHub repos, issues and workflows

    Not checked yet
    • Remote
    • Open source
    • 34.5K GitHub stars
    • Developer tools
  4. MCP toolkit giving coding agents symbol-level code retrieval and editing

    Runs locally
    • Open source
    • 31.5K GitHub stars
    • Developer tools
  5. XERJ

    MCP server

    Local search engine for AI agents with cited answers over your code and docs

    Runs locally
    • Open source
    • 19.2K GitHub stars
    • Knowledge and memory
  6. Ouroboros

    MCP server

    Spec-first layer that pins requirements and verifies results of AI coding agent work

    Runs locally
    • Open source
    • 6.4K GitHub stars
    • Developer tools