Hacker Next
Design Patterns for Securing LLM Agents Against Prompt Injectionssimonwillison.net
104 points by simonw 23 hours ago | 24 comments