Security Heuristics Regression: Excessive False Positives in version 5.5 xhigh

What version of the Codex App are you using (From “About Codex” dialog)?

26.513.31313 (2867)

What subscription do you have?

ChatGPT Pro 200USD/m

What platform is your computer?

MacOS

What issue are you seeing?

Version 5.5 incorrectly flags my work as network cracking. In reality, my operations are strictly local, primarily involving indexing and filtering a local database to match software metadata. These actions do not involve any network-based attacks or malicious exploits.

For my current project, I am developing software designed to identify which applications in a given directory are macOS apps. Once identified, it sorts out the Mac apps and deletes the non-Mac apps. Consequently, it needs to read various metadata within the applications, all of which are purely local behaviors. In the future, this program will be used to identify which GitHub repositories provide release links that download as Mac apps. All of this falls completely within the scope of normal usage. There is absolutely no behavior aimed at bypassing security mechanisms.

The current frequency of these alerts is intolerable and interferes with my daily workflow. I request that you investigate the changes made in the 5.5 update, address this over-sensitivity issue, and provide a method to suppress these false positives for my local development activities.

What steps can reproduce the bug?

Feedback ID: 019e20fb-2585-70c2-9205-8e76876ade32

What is the expected behavior?

_No response_

Additional information

_No response_