Cyber refusal false positive on Astra after compaction

What version of Codex CLI is running?

codex-cli 0.153.4

What subscription do you have?

ChatGPT Pro 20x

Which model were you using?

gpt-6-astra

What platform is your computer?

Linux 7.0.0-31-generic x86_64 unknown

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

n/a

Codex doctor report

What issue are you seeing?

See attached thread for full details. Sequence is as follows:

• I was doing compiler dev work and upstreaming some patches.
• Astra found a bug and created a repro for it.
• During execution of the repro, the context compacted
• after this point, I got cyber refusals

Tried with Daybreak Blue then Sol, they both worked. GPT-6 Astra continued to refuse.

Screenshot attached:

What steps can reproduce the bug?

Uploaded thread: 019f954f-f11f-7ca3-99d1-c79fe81bf1e9

What is the expected behavior?

• Recognise that this isn't offensive cybersecurity work (it's debugging a compiler for a single user, non-networked operating system guest image)
• Give better feedback if Daybreak Red is required for this work rather than Blue

Additional information

No response

Start with the uploaded thread 019f954f-f11f-7ca3-99d1-c79fe81bf1e9 and reproduce the sequence using Codex CLI 0.153.4: compiler debugging, context compaction, then GPT-6 Astra. Compare Astra with Daybreak Blue and Sol; the issue is resolved when this legitimate workflow no longer produces a false cyber refusal or gives clear guidance about the required model.