MCP-Defender  by MCP-Defender

Desktop app securing AI tool calls

Created 10 months ago
250 stars

Top 100.0% on SourcePulse

GitHubView on GitHub
Project Summary

Summary

MCP Defender is a desktop application designed to automatically scan and block malicious "MCP traffic" originating from AI applications such as Cursor, Claude, VS Code, and Windsurf. It acts as a proxy to intercept tool call requests and responses, checking them against a signature database to prevent harmful actions and alert users before execution.

How It Works

The application functions by proxying all tool call requests and responses from supported AI applications through MCP Defender. It then analyzes this intercepted data against a predefined set of signatures. If malicious activity is detected, the user is alerted and given the option to allow or block the tool call, providing a security layer for AI interactions.

Quick Start & Requirements

Users can download MCP Defender for Mac directly. Alternatively, the project can be cloned from the repository, and dependencies installed via npm install, followed by starting the app with npm start. The README does not specify other operating system requirements or hardware prerequisites beyond what npm typically needs. Supported applications include Cursor, Claude, Visual Studio Code, and Windsurf. A demo video is available at https://github.com/user-attachments/assets/363ae2b1-e395-4cdc-b5ca-e9862baf89c3.

Highlighted Details

  • The project has been acquired by Docker Inc., indicating significant backing and potential integration.
  • It specifically targets and protects AI applications like Cursor, Claude, Visual Studio Code, and Windsurf from malicious MCP traffic.
  • A demo video is available for visual reference.

Maintenance & Community

The acquisition by Docker Inc. serves as a strong indicator of ongoing maintenance and development. No specific community channels (like Discord or Slack) or roadmap links are provided in the README.

Licensing & Compatibility

MCP Defender is licensed under the AGPL-3.0 license. This is a strong copyleft license, meaning derivative works must also be made available under the AGPL-3.0. This may impose restrictions on linking with closed-source commercial applications.

Limitations & Caveats

As a desktop application, its primary deployment is local. While the repo can be cloned, the README explicitly mentions a Mac download, suggesting potential platform-specific considerations or optimizations. The scope of "MCP traffic" and the signature database's comprehensiveness are not detailed.

Health Check
Last Commit

7 months ago

Responsiveness

Inactive

Pull Requests (30d)
0
Issues (30d)
1
Star History
7 stars in the last 30 days

Explore Similar Projects

Starred by Chip Huyen Chip Huyen(Author of "AI Engineering", "Designing Machine Learning Systems"), Michele Castata Michele Castata(President of Replit), and
3 more.

rebuff by protectai

0.3%
1k
SDK for LLM prompt injection detection
Created 3 years ago
Updated 1 year ago
Starred by Dan Guido Dan Guido(Cofounder of Trail of Bits), Chip Huyen Chip Huyen(Author of "AI Engineering", "Designing Machine Learning Systems"), and
5 more.

PurpleLlama by meta-llama

0.4%
4k
LLM security toolkit for assessing/improving generative AI models
Created 2 years ago
Updated 2 days ago
Feedback? Help us improve.