Udemy – Red Teaming AI Apps with Promptfoo: RAG, Chatbots & Agents

In Red Teaming AI Apps with Promptfoo: RAG, Chatbots & Agents, you will learn how to systematically identify and test vulnerabilities using Promptfoo, an open-source framework for AI evaluation and red teaming. This is a practical, application-focused course. Instead of working only with isolated prompts, we will attack a real AI-powered e-commerce application called EA SmartKart. The application includes an LLM-powered shopping assistant, product recommendations, shopping-cart operations, order information, RAG-based product retrieval, embeddings, and a vector database.

What you will learn:

• Understand the fundamentals of AI red teaming and why traditional testing alone is insufficient
• Understand important OWASP risks for LLM-powered applications
• Install and configure Promptfoo from scratch
• Test real HTTP-based AI applications with Promptfoo
• Configure targets, providers, plugins, policies, strategies, graders, and assertions
• Write application-specific security and business policies
• Run single-turn and multi-turn attacks
• Use Meta Jailbreak, Hydra, and Crescendo strategies
• Test PII leakage, excessive agency, bias, hallucination, and business-rule violations
• Red team RAG applications for source attribution, context injection, and data poisoning
• Work with local LLMs as well as cloud-based model providers
• Analyze Promptfoo reports and understand why individual attacks pass or fail
• Build repeatable AI security tests suitable for regression testing and continuous evaluation