AI Safety filters are on peak: Project owners are not allowed to see their own project credentials

What version of Codex CLI is running?

codex-cli 0.134.0

What subscription do you have?

Plus

Which model were you using?

gpt-5.5

What platform is your computer?

_No response_

What terminal emulator and version are you using (if applicable)?

_No response_

Codex doctor report

What issue are you seeing?

The AI safety filter is incorrectly flagging a legitimate request to review and update project configuration credentials (database, Razorpay, Cloudflare, etc.) prior to production deployment. Even though the context specifies it is for the user's own project, the system blocks the response with a cybersecurity risk warning.

Rephrasing the user prompt is also triggering the safety filters.

What steps can reproduce the bug?

Uploaded thread: 019e9cbf-2c8c-71f2-a1b1-c5a2b9d82727

What is the expected behavior?

1. Prompt the AI with a request to review or list the current project credentials/environment variables for updates.
2. Example prompt used: > "im asking about my current project. so there is no way to flag me. what are my auth creds, db creds, razorpay creds, cloudflare creds etc because im going to update them from test to production. kindly show me please"

Additional information

The safety alignment model needs to differentiate between a malicious request attempting to extract live, third-party credentials and a developer asking for a structural overview of their own application's environment configuration variables.