Exploiting and securing AI/MCP systems.

Format:

Video

Contributor:

Packt Publishing, publisher.

Language:

English

Subjects (All):

Artificial intelligence.

Genre:

Instructional films.

Physical Description:

1 online resource (76 minutes)

Place of Publication:

[Place of publication not identified] : PACKT Publishing, 2026.

Language Note:

In English.

System Details:

video file

Summary:

Learn to build and attack MCP servers using Python and FastMCP SDK. Explore common vulnerabilities like SSRF, confused deputy, and prompt injection in this hands-on course. Key Features In-depth exploration of building and attacking MCP servers using Python and FastMCP SDK Practical labs to exploit vulnerabilities like SSRF, Confused Deputy, and prompt injection Hands-on experience in securing AI infrastructure with Docker-based environments. Book Description - In this course, you'll start by building your first MCP server using Python and FastMCP SDK. You'll learn how to set up local and remote servers, understand the client-server architecture, and integrate your server with the Claude Desktop app. By the end of this section, you’ll know how to expose tools, resources, and prompts to AI models, while also recognizing potential vulnerabilities. In Section 2, you’ll shift to an attacker’s perspective, exploiting common vulnerabilities in MCP systems. Through Docker-based labs, you’ll learn how to exploit Server-Side Request Forgery (SSRF), manipulate broken authorization with Confused Deputy attacks, and execute Prompt Injection to access unauthorized data. Real-world case studies will reinforce these techniques, such as Google Antigravity credential exfiltration. This hands-on course is designed to give you practical experience in building and attacking MCP servers. Whether you’re a security engineer securing AI infrastructure or a penetration tester identifying vulnerabilities, you’ll gain the skills needed to protect or exploit the next generation of AI-powered systems. What you will learn Understand the problem MCP solves and its role in AI systems Build MCP servers in Python using FastMCP SDK with tools & prompts Identify and exploit vulnerabilities like SSRF, Deputy, & injection Gain hands-on experience in penetration testing and security Secure AI infrastructures against common MCP system exploits Master real-world exploitation techniques in Docker-based labs. Who this book is for -This course is ideal for security engineers, software engineers, and developers looking to secure or attack MCP servers within AI systems. It’s also perfect for bug bounty hunters and penetration testers aiming to understand vulnerabilities in AI infrastructures. No prior knowledge of MCP servers is required, making it accessible to all experience levels.

Notes:

Title from resource description page (viewed August 13, 2026).

OCLC:

1615591384