Agentic Security
Not Started0 / 10 complete
Open a video or resource link, watch it, then check its box to track your progress and unlock what comes next.
10 lessons3 levels1 capstone project
Secure AI agents against prompt injection, excessive agency, unsafe tool use, identity and permission abuse, data exfiltration, memory poisoning, and tool/MCP supply-chain risk using least privilege, isolation, approvals, policy controls, evaluation, monitoring, and red teaming.
BASIC
- What Is a Prompt Injection Attack? — IBM Technology (8 min)
Build the threat model for malicious instructions, indirect prompt injection, and why untrusted context must be treated as hostile. - Prompt Injection Explained | The Biggest Security Risk in AI Agents — JetBrains / PyCharm (12 min selected)
Connect prompt injection to tool-using agents, unsafe actions, data access, and practical defensive patterns. See how injection can - The Promptware Kill Chain: How Prompt Injection Becomes AI Malware — IBM Technology (15 min selected)
progress from model manipulation to tool abuse, persistence, and compromise across agentic workflows.
INTERMEDIATE
- Securing AI agents on Azure — Microsoft Reactor
A deeper treatment of the same skill; use it to move from concept to working patterns. - Build Hour: Agentic Tool Calling — OpenAI
Apply Agentic Security through the adjacent capability: Agentic Tools Platform. - Orchestrate Your Agents with Agent Framework — Microsoft
Apply Agentic Security through the adjacent capability: AI Orchestration.
ADVANCED
- Beginner Automation Tutorial — n8n
Extend Agentic Security into senior/production practice through Workflow Automation. - Evaluate a RAG System Step by Step — RAGAS
Extend Agentic Security into senior/production practice through AI Evals & Monitoring. - Multi-Agent Swarms with LangGraph — LangChain
Extend Agentic Security into senior/production practice through AI Orchestration.