Agent Safety and Prompt Injection Guardrails Collection Curated papers, models, datasets, and demos for AI-agent runtime safety, prompt injection, MCP security, and tool-call guardrails. • 8 items • Updated about 20 hours ago • 1