Multi-Agent Pipeline for Protecting LLMs from Prompt Injection

An analysis of a multi-agent defense architecture that reduces prompt injection attack success by separating roles between agents.

15 October 2025