Skip to content

Learn · Track

Agent security

How agents get attacked through the text they read and the tools they call, and the layered controls that limit the damage, including how to keep personal data and secrets out of prompts, logs and traces. You should know what an agent and a tool call are; the agents track covers both.

Guides in this track

Intermediate6 guides, in reading order

  1. Prompt injection

    Direct and indirect prompt injection explained with agent examples, why filters alone fail, and the layered defenses OWASP and recent research recommend.

    9 min read

  2. OWASP Top 10 for LLM apps

    The ten OWASP LLM risks for 2025, one by one, with what each looks like in an agent that calls tools and a concrete control for each one.

    10 min read

  3. The lethal trifecta

    Why an agent with private data, untrusted content and a way to send data out is one prompt injection from a leak, and how to break the combination.

    9 min read

  4. MCP security risks

    The main MCP security risks from the spec and OWASP: tool poisoning, rug pulls, shadowing, confused deputy, token passthrough, SSRF and their controls.

    11 min read

  5. Least privilege for agents

    Why AI agents drift to more access than they need, the layers to narrow (servers, tools, arguments, data, people, keys) and how to keep them narrow.

    8 min read

  6. Sensitive data in prompts

    Where personal data and secrets leak in an AI app (prompts, tool arguments, logs, answers) and how masking, tokens and a controlled reveal contain them.

    9 min read

Before and after this track

This website (not the L2M product) uses Google Analytics, only after you accept cookies. Cookie settings