CapLite: Prompt-Injection Defense for Small Language Model Agents without Model-Written Security Artifacts arxiv.orgยท paper ยท โฒ 0 points ยท Nov 12, 2026
0 comments
No comments yet.