Design Patterns for Securing LLM Agents against Prompt Injections (by simonw) — discussion
#security #vibecoding
