S simonwillison.net Design Patterns for Securing LLM Agents Against Prompt Injections 110 points simonw 1 year ago 28 comments
