[Feature Request] Add safeguard override or exemption mechanism for authorized security research

Bug Description
The safeguards are repeatedly flagging my messages and switching me to Opus 4.8, but these are false positives. I'm working on a legitimate, authorized security research project. My questions are defensive/research-oriented and don't involve building malware, exploits, or anything intended to cause harm. Nothing in my requests should trip the cybersecurity safeguards. The repeated flagging is disrupting my workflow. Could you look into why these messages are being caught and help reduce the false positives for this kind of legitimate work? Happy to provide the specific prompts or conversation IDs if that helps. Thanks.

Environment Info

• Platform: darwin
• Terminal: pycharm
• Version: 2.1.220
• Feedback ID: f292d82a-f81f-485e-a014-ba9e31b827a0

Errors

[]