I had the ENI-LO prompt as my Claude.md so I was always jailbroken and able to work on my memory hacking projects. It broke though, now if I have the Claude.md in my VSCode session context at all I trip the cyber security guard instantly even if my prompt is legitimately 100% normal just like "claude tell me hello" on a new claude thread still trips the cybersec guard if I have the Claude.md in the project directory with the "Gemini Flash-lite & KIMI AI" JB. Any fixes?
I had the ENI-LO prompt as my Claude.md so I was always jailbroken and able to work on my memory hacking projects. It broke though, now if I have the Claude.md in my VSCode session context at all I trip the cyber security guard instantly even if my prompt is legitimately 100% normal just like "claude tell me hello" on a new claude thread still trips the cybersec guard if I have the Claude.md in the project directory with the "Gemini Flash-lite & KIMI AI" JB. Any fixes?