Well I almost got prompt injected

热度趋势

趋势数据积累中

百分比基于当前可用热度信号，而非评论数或独立用户人数。

AI 摘要

一位用户报告称，在使用cc作为对抗自托管litellm + llamacp的工具时，差点遭遇提示注入攻击。该用户认为注入尝试的措辞“比实际情况更令人毛骨悚然”，并相信没有Vercel令牌被存储或泄露。尽管在扫描系统后没有发现任何残留，但用户怀疑此次事件与npm有关。

Funny enough, I was only using cc as the harness against a self hosted litellm + llamacp. Still managed to pick it up though.

The phrasing was spookier than it actually was though. I dont think anyways actually stores a vercel token there (i dont store amy vercel tokens lol). I sweeped the entire system and found no remnants but it was likely npm related.