AgentStack
Browse Sign in
Browse Why AgentStack Sell Docs
Sign in
MCP verified MIT Self-run

E2b Sandbox Mcp

mcp-a2z2k26-e2b-sandbox-mcp · by a2z2k26

MCP server for orchestrating E2B cloud sandboxes and parallel agent work.

No reviews yet
0 installs
28 views
0.0% view→install

Install

$ agentstack add mcp-a2z2k26-e2b-sandbox-mcp

✓ scanned · ✓ verified, works with Claude Code, Cursor, and more.

Security review

✓ Passed

No issues found. Passed automated security review. · v0.1.0 How review works →

  • Prompt-injection patterns
  • Secret / credential exfiltration
  • Dangerous shell & filesystem operations
  • Untrusted network calls
  • Known-malicious package signatures

What it can access

  • Network access No
  • Filesystem access No
  • Shell / process execution No
  • Environment & secrets Used
  • Dynamic code execution No

From automated source analysis of v0.1.0. “Used” means the capability is present in the source — more access means more to trust, not that it’s unsafe.

View the full security report →

Verified badge

Passed review? Show it. Paste this badge into your README, it links to the public security report.

AgentStack Verified badge Links to your public security report.
[![AgentStack Verified](https://agentstack.voostack.com/badges/verified.svg)](https://agentstack.voostack.com/security/report/mcp-a2z2k26-e2b-sandbox-mcp)

Reliability & compatibility

Security review passed
0 installs to date
no reviews yet
3mo ago

Declared compatibility

Claude CodeClaude DesktopCursorWindsurf

Compatibility is declared by the source manifest. End-to-end runtime verification is coming, see below.

Preview Execution monitoring

We're building live execution health for every listing: tool-call success rate, median latency, uptime, and last-checked timestamps, measured, not self-reported. It isn't live yet, so we don't show numbers we can't stand behind.

How agent discovery & health will work →
Are you the author of E2b Sandbox Mcp? Claim this listing to set pricing, connect Stripe payouts, and keep 70% of every sale.
Sign up to claim

About

 _____ ____  ____    ____    _    _   _ ____  ____   _____  __
| ____|___ \| __ )  / ___|  / \  | \ | |  _ \| __ ) / _ \ \/ /
|  _|   __) |  _ \  \___ \ / _ \ |  \| | | | |  _ \| | | \  /
| |___ / __/| |_) |  ___) / ___ \| |\  | |_| | |_) | |_| /  \
|_____|_____|____/  |____/_/   \_\_| \_|____/|____/ \___/_/\_\

 __  __  ____ ____
|  \/  |/ ___|  _ \
| |\/| | |   | |_) |
| |  | | |___|  __/
|_|  |_|\____|_|

[](https://www.typescriptlang.org)

Orchestrate sandboxes from any model. Parallel agents. Drop-in MCP. E2B Sandbox MCP is a Model Context Protocol server that lets any MCP-compatible AI client spin up isolated cloud sandboxes through E2B, run code in them, and coordinate parallel agents — all from inside a conversation. Built for the case where the model itself is the orchestrator. ###


Sandbox Lifecycle ###

  1. Initialize: create a sandbox with a chosen template (Python, Node, Go, Rust, base).
  2. Operate: read and write files, run shell commands, stream output.
  3. Tear down: pause, resume, or kill the sandbox when work is complete.

Multi-Agent Orchestration ###

  • Parallel sandboxes: dispatch multiple agents across isolated environments.
  • Dependency-aware scheduling: analyze_dependencies resolves build order automatically.
  • Resource allocation: balanced, max-speed, and cost-aware strategies.

Model-Agnostic ###

  • Any MCP client — Claude Desktop, Claude Code, Cursor, Continue, custom.
  • Stdio transport with no model dependency.
  • Your client picks the model. The server only exposes the tools.

Tool Inventory ###

| Category | Tools | What you can do | |----------|-------|------------------| | Lifecycle | 5 | Create, connect, pause, resume, kill sandboxes | | File ops | 10 | Read, write, list, copy, move, sync inside sandboxes | | Commands | 1 | Execute shell commands with streaming + exit codes | | Orchestration | 8 | Plan allocation, spawn agents, monitor, handle events | | Dependencies | 1 | Analyze a GitHub repo's dependency graph |

24 tools total. Full reference: [docs/MCP_TOOLS_REFERENCE.md](docs/MCPTOOLSREFERENCE.md).

Installation ###

(requires Node.js 18+, an E2B account, and an MCP-compatible client)

# Clone and install
git clone https://github.com/a2z2k26/e2b-sandbox-mcp.git
cd e2b-sandbox-mcp
npm install
npm run build

Configuration ###

Add to Claude Desktop (~/Library/Application Support/Claude/claude_desktop_config.json on macOS):

{
  "mcpServers": {
    "bumba-sandbox": {
      "command": "node",
      "args": ["/absolute/path/to/e2b-sandbox-mcp/dist/mcp-servers/bumba-sandbox.js"],
      "env": {
        "E2B_API_KEY": "",
        "GITHUB_TOKEN": "",
        "GITHUB_REPO_OWNER": "your-username",
        "GITHUB_REPO_NAME": "your-repo"
      }
    }
  }
}

Cursor, Continue, and other MCP clients use the same shape — point them at dist/mcp-servers/bumba-sandbox.js with the same env block.

Environment Setup ###

Copy the template and fill in your keys:

cp .env.example .env
# Required
E2B_API_KEY=
GITHUB_TOKEN=
GITHUB_REPO_OWNER=your-github-username
GITHUB_REPO_NAME=your-repository-name

# Optional
# E2B_API_URL=https://api.e2b.dev
# GITHUB_API_URL=https://github.your-company.com/api/v3
# LOG_LEVEL=info
# MAX_CONCURRENT_SANDBOXES=5

First Sandbox ###

Once connected, ask your model:

Spin up a Python sandbox, install pandas,
and run analyses/q1.py against data.csv.
Report back with the results.

The model invokes sandbox_initfile_writecommand_executesandbox_kill to handle the full lifecycle.

Parallel Agents ###

Implement issues #12, #13, and #14 in parallel sandboxes.
Stop early if any of them fail tests.

analyze_dependencies and spawn_parallel_agents handle dependency-aware scheduling automatically. Use set_orchestration_strategy to switch between balanced, max-speed, and cost-aware.


Architecture ###

   MCP client                 MCP server                External
  ┌───────────┐   stdio    ┌─────────────────┐
  │  Claude / │ ─────────► │  bumba-sandbox  │ ──► E2B Cloud Sandboxes
  │  Cursor / │            │      -mcp       │ ──► GitHub API
  │  custom   │ ◄───────── │                 │ ──► Local state / logs
  └───────────┘            └─────────────────┘

Deep dive: [docs/BUMBA_SANDBOX_ARCHITECTURE.md](docs/BUMBASANDBOXARCHITECTURE.md).

Development ###

npm run build       # compile TypeScript
npm run watch       # rebuild on change
npm test            # run jest test suite (111 tests)
npm run lint        # eslint
npm run type-check  # tsc --noEmit

Documentation ###

  • [docs/QUICK_START_GUIDE.md](docs/QUICKSTARTGUIDE.md) — 10-minute end-to-end walkthrough
  • [docs/SETUP.md](docs/SETUP.md) — full setup with troubleshooting
  • [docs/MCP_TOOLS_REFERENCE.md](docs/MCPTOOLSREFERENCE.md) — every tool, every parameter
  • [docs/BUMBA_SANDBOX_ARCHITECTURE.md](docs/BUMBASANDBOXARCHITECTURE.md) — system design

License ###

[MIT](LICENSE) — Andrew Zellinger.

Built on E2B, Model Context Protocol, and Octokit.

Contributing ###

Issues and PRs welcome. For larger changes, please open an issue first to discuss the approach.

Source & license

This open-source MCP server is cataloged on AgentStack and links to its original source — we do not rehost the code.

Install and usage instructions live in the source repository linked above.

Reviews

No reviews yet, be the first.

Versions

  • v0.1.0 Imported from the upstream source.