RCCE Course
Course #473

Essentials of Prompt injection

📊 Level: Beginner
⏱️ Duration: 2 Days
🏷️ Track: AI Security
📋 Prerequisites: None
🖥️ Mode: Online Instructor-Led
📝 Course Description

RCCE students will learn prompt injection attacks against AI/LLM systems including direct prompt injection, indirect prompt injection, jailbreaking techniques, and prompt leaking. RCCE students will learn to identify prompt injection vulnerabilities in AI-powered applications, execute prompt injection attacks in controlled environments to demonstrate data extraction, instruction override, and unintended actions, implement defensive measures including input sanitization, system prompt hardening, output filtering, and architectural separation of trusted and untrusted content, monitor AI systems for prompt injection attempts, and develop incident response procedures for compromised AI systems. This essentials course covers the core knowledge needed to operate competently in this domain. Starting from foundational concepts, RCCE students will learn the fundamental concepts, terminology, risks, and defenses that form the foundation for all further study and professional practice. Students build a solid knowledge base that prepares them for more advanced courses and real-world security responsibilities.

🎯 Target Audience
  • Security Engineers building defensive controls
  • Security Analysts and Blue Team members
  • Systems Administrators with security responsibilities
  • GRC and Risk Professionals supporting controls
  • Professionals implementing Essentials of Prompt injection
🧠 What You Will Learn
  • Explain Course Overview fundamentals
  • Execute hands-on tasks for what you will learn
  • Execute hands-on tasks for course structure
  • Execute hands-on tasks for learning objectives
  • Integrate privilege controls with identity providers and SIEM telemetry, including Perform controlled injection.
  • Design a scalable privilege management architecture with policy and enforcement
  • Execute hands-on tasks for core concepts
  • Execute hands-on tasks for key characteristics — covering Neural networks trained on text data, Follow instructions in natural language.
  • Execute hands-on tasks for user input
  • Execute hands-on tasks for input layer
  • Execute hands-on tasks for processing layer
📚 Course Outline
Module 01Course Overview
Module 02What You Will Learn
Module 03Course Structure
Module 04Learning Objectives
Module 05Assess LLM integration risks
Module 06What Are Large Language Models
Module 07Core Concepts
Module 08Key Characteristics
Module 09User Input
Module 10→ Tokenization → Context Window → Model Inference → Output Generation
Module 11Input Layer
Module 12Processing Layer
Module 13Output Layer
Module 14User Interface
🧪 Lab Details

All hands-on labs run on Rocheston Rose X OS. Students practice essentials of prompt injection by implementing the controls discussed in class, with a focus on real-world deployment, monitoring, and validation.

  • Lab 1: Explain Course Overview fundamentals
  • Lab 2: Execute hands-on tasks for what you will learn
  • Lab 3: Execute hands-on tasks for course structure
  • Lab 4: Execute hands-on tasks for learning objectives
  • Lab 5: Integrate privilege controls with identity providers and SIEM telemetry
📊 Skill Level
Beginner
Beginner Intermediate Advanced Expert
Duration
2 Days
🎓
Certificate
Completion
🖥️
Lab Platform
Rose X OS
👨‍🏫
Mode of Training
Online Instructor-Led
🔥
Platform
Zelfire
🐦‍⬛
Cyber Range
Raven
📓
Study Material
CyberNotes
🏆 Certificate

Upon successful completion of this course, students will receive an official RCCE Course Completion Certificate for Essentials of Prompt injection, verifiable through the Rocheston certification portal.

🔑 Student Access & Materials
  • Full access to all course materials and slide decks
  • Hands-on lab access on Rocheston Rose X OS environment
  • Access to Rocheston CyberNotes
  • Access to Rocheston Zelfire — EDR/XDR SIEM platform
  • Access to Rocheston Raven — online cyber range exercise platform
  • Access to Rocheston Vulnerability Vines AI