Sandbox
@MCP-Defender/MCP-Defender

Desktop proxy for MCP traffic security

MCP Defender watches MCP tool call requests and responses from supported AI apps, then checks them against stored signatures. When it spots something risky, it interrupts the flow and asks whether to allow or block the call.

257 starsโ€ข42 forksโ€ขTypeScriptโ€ขUpdated 3mo ago
Who it's for

Builders who want their agent traffic inspected and filtered before tool calls reach external services.

What it delivers

You can keep using your AI apps while getting a prompt when an MCP call looks unsafe.

What it does

Proxy MCP traffic

Routes MCP tool call requests and responses through the desktop app instead of sending them straight through.

Signature-based detection

Compares intercepted MCP data against the bundled signature sets to find harmful patterns.

Allow or block prompts

Alerts you when suspicious traffic is detected and lets you decide whether the call should continue.

Support for multiple AI apps

Works with Cursor, Claude, Visual Studio Code, and Windsurf according to the README.

How to get it

  1. 1Alternatively you can clone the git repo, and run it as follows
    # Install dependencies
    npm install
    
    # Start app
    npm start

README

Shows a black knight in light color mode and a white knight one in dark color mode.Shows a black knight in light color mode and a white knight one in dark color mode.

MCP Defender

Automatically protects MCP traffic in AI apps

GitHub License

MCP Defender has been acquired by Docker Inc.

๐Ÿ›ก๏ธ MCP Defender is a desktop app that protects AI apps like Cursor from a variety of attacks.

๐Ÿšฆ All MCP tool call requests and responses from AI apps are automatically proxied through MCP Defender.

๐Ÿ”Ž The intercepted data is then checked against a set of signatures.

๐Ÿ” If anything harmful is detected, MCP Defender alerts you and asks if you want to allow or block the tool call.

Demos

https://github.com/user-attachments/assets/363ae2b1-e395-4cdc-b5ca-e9862baf89c3

Quick Start

Download MCP Defender for Mac

Alternatively you can clone the git repo, and run it as follows:

# Install dependencies
npm install

# Start app
npm start

Which apps are automatically protected?

MCP Defender protects Cursor, Claude, Visual Studio Code and Windsurf.

License

MCP Defender is licensed under the AGPL-3.0 license. For more details, see the LICENSE.

Files in the repo

Repository payloadโ€ข27 top-level entries
  • .cursor
  • .github
  • .vscode
  • signatures
  • src
  • tests
  • .eslintrc.json
  • .gitignore
  • DEVELOPMENT.md
  • forge.config.ts
  • forge.env.d.ts
  • index.html
  • LICENSE
  • mcp-defender-overview.mdc
  • package-lock.json
  • package.json
  • postcss.config.js
  • PROXY.md
  • publish.sh
  • README.md
  • tailwind.config.js
  • tsconfig.cli.json
  • tsconfig.json
  • vite.main.config.ts
  • vite.preload.config.ts
  • vite.renderer.config.ts
  • vite.utility.config.ts

Discussion (0)

Ask about usage, or say what you built with it

Sign in to join the discussion.

No comments yet. Be the first to say what this is good for.

More tools

JuliusBrussee/
caveman

๐Ÿชจ why use many token when few token do trick โ€” Claude Code skill that cuts 65% of tokens by talking like caveman

105k
1 add
MemPalace/
mempalace

The best-benchmarked open-source AI memory system. And it's free.

59k
stablyai/
orca

Orca is the ADE for working with a fleet of parallel agents. Run any coding agent with your own subscription. Available on desktop, mobile and remote runtime.

66k

A cross-platform desktop All-in-One assistant for Claude Code, Codex, OpenCode, OpenClaw, Grok Build & Hermes Agent. Only official website: ccswitch.io

132k

Never stop coding. Free MIT AI gateway: one endpoint, 352 providers (150+ free), 1200+ models Kimi, Claude, GPT, Gemini, GLM, DeepSeek, MiniMax. Works with Claude Code, Codex, Cursor, OpenCode, Cline & Copilot. Quota-aware auto-fallback, RTK+Caveman compression saves 15-95% tokens, MCP/A2A, Desktop/PWA. Built by 550+ contributors

64k
headroomlabs-ai/
headroom

Compress tool outputs, logs, files, and RAG chunks before they reach the LLM. 20% fewer tokens for coding agents, 60-95% fewer tokens for JSON, same answers. Library, proxy, MCP server.

71k