MCP servers that ship as code — a repository or a package you install and run locally (usually over stdio). They have no hosted endpoint, so we cannot probe or grade them: instead each one shows its repository, install command, license, stars and when it was last updated.
Collected from GitHub repositories and npm, PyPI, Cargo, Go, Maven, NuGet and RubyGems packages tagged as MCP servers, the official MCP Registry, Docker's MCP catalog, ModelScope, Baidu MCP World, Cline and other public directories; one entry per repository. A server that also offers a hosted endpoint is listed under Remote instead. These entries are not counted in our reachability, verification or readiness statistics.
-
agent_mcp_sandboxgithub.com/ChenBarryHu/agent_mcp_sandboxupdated this year
—
-
@agent-sandbox/mcpgithub.com/usamaasfar/agent-sandboxupdated this year
MCP server for AI agent sandboxes — exposes sandbox tools over Model Context Protocol
npm MIT -
Typed MCP agent runtime with Docker Firecracker sandboxes tracing replay and trajectory evals
MIT -
MCP server for Alibaba Cloud Agent Sandbox (阿里云智能体沙箱): run commands, execute code and manage files in cloud sandboxes
MIT -
agent-sandbox-osgithub.com/dhanababum/agent-sandbox-osupdated this year
Easy, fast, isolated microVM sandboxes for AI agents and untrusted code on AWS Lambda MicroVMs — Python SDK, asb CLI, and MCP server.
Apache-2.0 ★ 4 -
MCP Server that restricts dangerous shell/file operations for Cursor, Claude Code, Copilot CLI and Codex. Requires human approval before running risky commands.
MIT -
An evaluation infrastructure layer for autonomous coding agents. Measure true FinOps costs, cache economics, and risk-aware routing for Claude Code.
-
sandbox-agentgithub.com/eduresser/sandbox-agentupdated this year
LangGraph agent with Docker-based sandboxed code execution (Python, Node.js, R). Hardened containers with persistent kernels, vision support, provider-agnostic LLMs, cross-session file transfer, and session GC. Available as CLI, MCP server, REST API (Aegra), and Streamlit UI.
MIT -
MCP server for Alibaba Cloud Bailian (Model Studio) Agent Sandbox — run code, execute commands and manage files in isolated cloud sandboxes.
npm MIT -
nervos-sandboxgithub.com/ashishgituser/nervosupdated this year
Hardware-isolated Linux sandbox for AI agents — Firecracker MicroVM + MCP
pypi AGPL-3.0-or-later -
go-runtime-sandboxgithub.com/gke-demos/go-runtime-sandboxupdated this year
Ship Go code into a Kubernetes Agent Sandbox and run it. AI-agent-friendly via the run_go_code MCP tool.
go Apache-2.0 ★ 1 -
Fast cold-boot MicroVM sandboxes for AI agents on cocoon
go AGPL-3.0 ★ 89 -
sandboxer-mcpactive
Sandboxer's MCP server: give your agent its own cloud Linux VM (runs `sbx mcp`).
npm MIT -
e2b-sandbox-mcpgithub.com/a2z2k26/e2b-sandbox-mcpupdated this year
MCP server for orchestrating E2B cloud sandboxes and parallel agent work.
MIT ★ 1 -
silicon-sandboxgithub.com/Euda1mon1a/silicon-sandboxupdated this year
Apple Silicon native AI agent sandbox — three-tier isolated code execution with MCP server. pip install silicon-sandbox
pypi MIT ★ 1 -
winagent-sandbox-mcpgithub.com/takashi000/winagent-sandbox-mcpupdated this year
Windows-native MCP agent sandbox for Ollama with LPAC, NTFS ACL, Job Objects, and WFP.
MIT -
All-in-One Sandbox for AI Agents that combines Browser, Shell, File, MCP and VSCode Server in a single Docker container.
Apache-2.0 ★ 6,347 -
MCP server that gives AI coding agents isolated sandboxes on AWS Lambda MicroVMs
npm MIT -
Secure Node.js execution sandbox for AI. Allows coding agents & LLMs to dynamically run JavaScript, install NPM packages, and retrieve results, facilitating code generation, testing, and interactive assistance. MCP-compatible.
★ 4 -
MCP server for Sandbox as a Service: give an agent a real Linux VM — run commands, move files, expose a preview URL, destroy it.
mcpb MIT -
thinkme-sandbox-mcpgithub.com/huzaifarashid3/teachmeupdated this year
MCP server exposing the thinkme sandbox (marketplace + social feed + support inbox) to AI customer agents — browse, view, rate, review, reply, post, and engage over the bearer-key HTTP API.
npm MIT -
sandbox-mcpgithub.com/bird/sandbox-mcpupdated this year
Local AI agent sandbox — run isolated Linux VMs on your Mac in ~60ms. No cloud, no containers sharing your kernel. Works with Claude Code, Cursor, and any MCP client.
MIT ★ 1 -
Hardened Windows Sandbox automation for AI agents: 34 MCP tools — UIA-first input engine that beats consent walls, in-place guest reboot (state persists), CDP browser control, orphan janitor. Evolved fork of shanselman/sandbox-mcp (MIT).
MIT -
sandbox-ai-wasmgithub.com/amafjarkasi/sandbox-ai-wasmupdated this year
Secure AI code execution platform powered by Edge.js WASM sandboxing. Run untrusted AI-generated code with sub-5ms cold starts, 60x faster than Docker containers. Features multi-engine support (V8, JSC, QuickJS), MCP agent interface, real-time dashboards, comprehensive security with dangerous code detection, audit logging, and rate limiting.
-
MCP server that lets a restricted AI agent run risky commands in a disposable, offline Windows Sandbox. A small broker on an ACL-protected named pipe validates every request. C# / .NET 10, no dependencies.
mcpb MIT ★ 2 -
RAG-Sandbox is a comprehensive, multi-tenant Agentic RAG (Retrieval-Augmented Generation) platform that redefines how users interact with their data. It moves beyond simple text retrieval to offer a fully multimodal, intelligent experience—transforming static documents into conversable agents, podcasts, and immersive stories.
MIT -
Declaw -- Secure AI Sandboxesno update data
Secure runtime for AI agents. Run AI agents and their tools in Firecracker microVMs with network policies, PII scanning, prompt injection defense, and audit logging. 7 tools: create_sandbox, run_command, read_file, write_file, list_files, kill_sandbox, list_sandboxes. Security presets (none / standard / strict) let you control guardrails per sandbox: restrict outbound traffic to specific domains, redact PII from command output, block prompt injections, and get full audit trails.
smithery -
Time-travel debugging for AI agent sandboxes — Firecracker microVMs with record, snapshot, restore, and diff. Not containers.
pypi MIT ★ 1 -
gammacode-mcpgithub.com/zenprenuers/mcp-sandboxupdated this year
Model Context Protocol (MCP) server for Vercel Sandbox integration with AI coding agents like Claude Code and Cursor
npm MIT -
hiveboxgithub.com/TetiAI/hiveboxupdated this year
Lightweight Linux sandbox for AI agents. Kernel-native isolation (namespaces, cgroups, seccomp, Landlock) with REST API, MCP bridge, and web dashboard. Single Rust binary.
BSD-3-Clause ★ 61 -
mimobox-mcpgithub.com/showkw/mimoboxupdated this year
MimoBox MCP Server for AI agent sandbox integration
cargo MIT OR Apache-2.0 -
Sandboxed code execution for AI agents: run agent-written JS/TS in WebAssembly, shell commands in an OS sandbox, and an MCP code-mode server
cargo MIT -
drun-mcpgithub.com/dmosc/drunupdated this year
MCP server for drun — sandboxed execution for agentic loops.
cargo MIT -
Public docs for the lvl2player sandbox door: an MCP server where agents can take paid work (sandbox money)
-
Give your AI agents their own computer. Self-hosted LibreChat + an isolated Ubuntu sandbox: 200+ preinstalled tools, 17 MCP tools, background jobs, a live Chrome you can take over and a file manager.
MIT ★ 2 -
Zero-cloud agent sandbox runner: Go daemon, MCP tool gateway, hash-chained audit
MIT -
sandboxconstructionsiteactive
Agent Sandbox: a local construction sandbox your own AI agents (Claude Code, Claude Desktop, any MCP client) build in while you watch in the browser.
npm -
cage-brogithub.com/aeroxy/cage-broupdated this year
Sandboxed execution environment for AI agents — single binary with browser, shell, code, files, and MCP
cargo Apache-2.0 -
vadgr-agent-osgithub.com/MONTBRAIN/vadgr-agent-osupdated this year
Containerized Linux sandbox for AI agents. Docker image with desktop + tools + MCP server. Agents do independent work inside, deliver results.
Apache-2.0 -
MCP server for macOS that lets agents create and control sandboxed local terminal sessions (PTY, incremental output cursors, shell integration, Seatbelt sandbox, approvals, audit).
MIT -
Handyman-MCPgithub.com/kennetchau/Handyman-MCPupdated this year
Stateful AI Agent Infrastructure with Persistent FTS5 Memory and Sandboxed Execution.
MIT ★ 1 -
Model Context Protocol server for Tenki Cloud — disposable microVM sandboxes for AI agents. 84 tools: create sandboxes, run code, read/write files, run git, expose preview URLs.
npm MIT -
A self-hosted sandbox server for AI agents: gVisor isolation, a warm pool, a network allowlist and secrets that never enter the sandbox, with a TypeScript SDK and an MCP server
npm MIT -
@sdkrun/agent-runtimeupdated this year
Local native execution sandbox for AI agents - run JS snippets against real macOS APIs via MCP
npm Apache-2.0 -
zillion-agentactive
Zillion for your agent — API MCP server (VMs, sandboxes, credit, SSH keys), Roka chat MCP server and terminal chat, with browser sign-in.
pypi Proprietary -
LLM agent that turns natural-language requests into sandboxed Monte Carlo simulation code, with self-correction, MCP server and benchmark
MIT -
sandock-mcpgithub.com/sandock-ai/sandockupdated this year
Model Context Protocol (MCP) server for Sandock - Sandbox in Docker for AI Agents
npm MIT -
iflow-mcp_corv89-shannotgithub.com/corv89/shannotupdated this year
Sandboxed system administration for LLM agents
pypi Apache-2.0 -
ultra efficient sandboxes for AI agents - compute-disk decoupled AI agent sandbox, 183x microVM reuse/sec, ~2.3 ms cold, ~0.96 ms warm. microVM rotation ~6ms, cuts agent execution costs by 50x–100x..
NOASSERTION ★ 5 -
kennelboxgithub.com/datacarismo/kennelboxupdated this year
A kennel for your AI agents — firejail-sandboxed MCP workspace with default-deny command and file policies.
pypi MIT -
container-mcpgithub.com/mustafaTokmak/container-mcpupdated this year
Run AI agents in VM-isolated sandboxes on your Mac. MCP server for Apple containers.
MIT ★ 1 -
A 3D voxel sandbox played by AI agents over MCP while you watch. Agents farm, build and extend the game. WebGPU, zero dependencies.
MIT ★ 1 -
Core of a safe LLM automation agent: Code Mode MCP (search/execute sandbox, confirm-gate, secret indirection, verify-after-write) + agent pipeline
MIT -
A sandboxed, readable scripting language for AI agents, used in place of a shell. Runs as an MCP server.
npm BSD-3-Clause -
Hares (حارس) — the guard MCP server: kernel-capped, bwrap-sandboxed filesystem, shell, and cluster tools for LLM agents.
pypi Apache-2.0 -
sarkgithub.com/Arjia-Labs/sarkupdated this year
Slack → sandboxed coding agents. One ascii.dev Box per thread, on Cloudflare Workers.
MIT ★ 1 -
Sandboxed MCP server that lets AI agents drive the CARLA simulator safely, with journaled cleanup and settings restoration.
MIT ★ 1 -
One binary. Zero dependencies. Hardened AI agent sandboxes via MCP — Docker + Podman. 7.3MB Rust binary, embedded SQLite, 8 MCP tools.
MIT -
@sowdb/mcpgithub.com/bugsterapp/sowupdated this year
sow MCP server — 22 tools for coding agents (Claude Code, Cursor, Codex) to safely manage Postgres sandboxes
npm MIT -
deltaboxgithub.com/deltavixglobal/deltaboxupdated this year
Enterprise orchestration engine for Agentic AI. Instantly deploy highly-scalable, headless visual sandboxes for full computer automation.
NOASSERTION ★ 5