I Had Claude MCP Hack Me

Item: I Had Claude MCP Hack Me
Rating: 4
Author: PostSingularity Institute

by John Hammond

Applied Sciences & Engineering Computing & Cybersecurity URComputer security URHComputer fraud and hacking

🎙 John Hammond 👥 2.2M 📅 November 14, 2025 ⏱ 21 min 👁 38K 📄 tutorial 🧭 2026-08-16

Available in: English (current) Français

Keywords

MCPClaudeAtomic Red TeamAI hackingcybersecurity

Summary

In this video, John Hammond explores the intersection of AI and cybersecurity by demonstrating how Claude, an AI assistant, can be used to execute adversary emulation techniques via the Model Context Protocol (MCP). He introduces the Atomic Red Team MCP server, a tool developed by Cyberbuff that allows AI to interact with the Atomic Red Team library, which contains tests mapped to the MITRE ATT&CK framework. The video begins with an explanation of the MCP and its potential security implications, followed by a sponsored segment for Harmonic Security’s MCP Gateway. Hammond then proceeds to set up the MCP server in a Windows 11 virtual machine using Cursor, an AI-powered code editor. He demonstrates how Claude can query and execute various atomic tests, such as creating scheduled tasks, changing desktop wallpaper, and creating a Windows service, all without manual intervention. The video highlights the power and risks of allowing AI to execute code, emphasizing the need for security controls. Hammond also mentions Cyberbuff’s follow-up blog post about using Claude as a command and control server. The video concludes with a discussion on the future of AI in cybersecurity and encourages viewers to explore the tool responsibly.

195 words

Critical Evaluation

Value of the Information & Strength of the Argument

The video provides valuable insights into the practical application of AI in cybersecurity, specifically in adversary emulation and detection engineering. It demonstrates a real-world use case of MCP servers, which is a relatively new and evolving technology. The argumentation is solid, as Hammond clearly explains the steps and the reasoning behind using the MCP server, while also acknowledging the potential risks and ethical considerations. He emphasizes the importance of testing in controlled environments and obtaining proper authorization. The demonstration is compelling and effectively showcases the capabilities of the tool, making a strong case for its utility in speeding up threat intelligence workflows.

Scientific Rigor, Source Quality, Title Accuracy

The video is scientifically rigorous in its technical execution, with clear references to the original tools and blog posts by Cyberbuff. The sources cited include the Atomic Red Team GitHub repository, the MCP server repository, and Cyberbuff’s blog posts. The title accurately reflects the content, and the video does not overstate its claims. The presenter also mentions the sponsor, Harmonic Security, but this does not detract from the technical content. The video is well-structured and provides a balanced view of the benefits and risks of using AI in cybersecurity.

206 words

Title / Content Match

The title accurately reflects the content: the video demonstrates Claude (via MCP) executing Atomic Red Team tests on the presenter's system, effectively 'hacking' him in a controlled manner.

Quality & Reliability

8/10

The video is a practical demonstration of using the Atomic Red Team MCP server with Claude, based on the work of Cyberbuff. The content is technically accurate, well-explained, and includes references to the original tools and blog posts. The demonstration is conducted in a controlled environment (Windows VM) and the presenter emphasizes responsible use. However, the video is primarily a tutorial and does not provide deep scientific analysis or peer-reviewed sources.

Key Moments

Markers derived by PSI from the transcript: the creator did not define chapters.

Cited Sources

Atomic Red Team MCP Blog Post — Cyberbuff's blog post introducing the Atomic Red Team MCP server.
Atomic Red Team MCP - Claude Becomes the APT — Follow-up blog post on using Claude as an APT.
Claude Becomes the APT — Initial blog post on using Claude with Atomic Red Team.
Cyberbuff's GitHub — GitHub profile of the tool's developer.
Atomic Red Team MCP Repository — Source code for the MCP server.
Atomic Red Team Repository — Official Atomic Red Team library.

Concurring Sources

Atomic Red Team MCP Blog Post — The blog post provides detailed information on the MCP server, which aligns with the video's content.
Atomic Red Team MCP - Claude Becomes the APT — Follow-up post that expands on the capabilities demonstrated in the video.

External References

Contribution & Novelties

This video provides a practical demonstration of integrating AI with adversary emulation, showcasing a novel use of MCP servers to automate and accelerate threat intelligence workflows. It highlights the potential of AI to handle complex tasks such as querying and executing atomic tests, which traditionally require manual effort. The video also raises important security considerations regarding the use of AI in cybersecurity, emphasizing the need for proper controls and visibility.

Pour aller plus loin :

Model Context Protocol (MCP) — Official documentation on MCP, the protocol enabling AI to interact with external tools.
MITRE ATT&CK Framework — Comprehensive knowledge base of adversary tactics and techniques.
Atomic Red Team — Library of tests mapped to MITRE ATT&CK for adversary emulation.

119 words

Radar Profile

The radar profile shows high scores in quantity and quality of information, indicating a well-structured and informative video. The technical level is moderately high, suitable for an audience with some cybersecurity background. The overall reliability is strong, supported by credible sources and a clear demonstration.

Reliability 8/10

💬 Sur les 0 commentaires analysés, aucune tendance n'est disponible.