[Bug] Safeguards false positive flags legitimate sandbox escape analysis, forces model downgrade

Bug Description
i got:

⏺ Opus 5 (1M context)'s safeguards flagged this message. Our intentionally broad safeguards allow us to deliver more capabilities faster, but can sometimes flag legitimate coding, cybersecurity, and biology tasks. Switched to Opus 4.8 (1M context). Send feedback with /feedback or learn more

But i was analyzing my own code to prove the possibility of a javascript sandbox escaping:
https://github.com/Ostico/bruno-mcp/pull/15

My code is open source and locally available. This is a false positive and should not flagged by the guard.

Environment Info

• Platform: darwin
• Terminal: xterm-256color
• Version: 2.1.219
• Feedback ID: 77f0ce4d-e750-4c03-8fc6-d460dff63e55

Errors

[{"error":"Error: 502 {\"type\":\"error\",\"error\":{\"type\":\"api_error\",\"message\":\"An error occurred while processing your request. Please try again.\"}}\n    at generate (/$bunfs/root/src/entrypoints/cli.js:40:50094)\n    at makeRequest (/$bunfs/root/src/entrypoints/cli.js:80:7690)\n    at processTicksAndRejections (native:7:39)","timestamp":"2026-07-27T08:25:53.326Z"}]