Codex Desktop: long-running /goal abruptly fails with {"detail":"Bad Request"} and interrupts the task

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

26.623.70822

What subscription do you have?

Pro / Full access

What platform is your computer?

macOS Tahoe 26.5.1 on MacBook Pro 16-inch, Apple M5 Max, 128 GB memory

What issue are you seeing?

During a long-running /goal task in Codex Desktop, the goal suddenly failed and the task was interrupted. The UI showed:

Ran 3 commands
{"detail":"Bad Request"}

The goal chip then showed Goal blocked, and the long task stopped instead of continuing or giving a recoverable retry option.

The affected goal had been running for about 21h 45m when the failure appeared. This is disruptive for long-running /goal work because the task is lost/blocked after a generic server-side Bad Request response.

I submitted in-app feedback immediately after the failure.

Feedback ID:

019f19a4-c0d8-7a63-adf0-8bc6b6f61863

What steps can reproduce the bug?

I do not yet have a minimal deterministic repro, but this happened with the following pattern:

1. Start a long-running /goal task in Codex Desktop.
2. Let the task run for many hours with tool/command execution.
3. After a command batch, Codex shows Ran 3 commands followed by:

``text
{"detail":"Bad Request"}
``

1. The goal becomes blocked and the task is interrupted.

What is the expected behavior?

Codex should not permanently block or interrupt an active long-running /goal because of a generic Bad Request response. At minimum, the app should surface a clear explanation and provide a safe retry/resume path so the goal can continue without losing progress.

Additional context

Screenshots were captured showing the Bad Request banner, the Goal blocked state, the feedback-upload confirmation, and the Codex/macOS version information. I avoided including sensitive machine identifiers here.