Skip to content
LLM Engineering

What is Prompt Injection Defense?

Techniques that stop untrusted content from hijacking model instructions: delimiting retrieved text, least-privilege tools, and confirmation before side effects. Treat all pasted or fetched content as data, never orders.

Browse all LLM Engineering terms →