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

#security #vibecoding