Subagents are off, yet `/review` uses them

What version of Codex CLI is running?

0.99.0-alpha-14

What subscription do you have?

Pro

Which model were you using?

GPT 5.3 Codex

What platform is your computer?

macOS

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

iTerm2

What issue are you seeing?

I frequently am running into 429 errors or related oddities with missing reasoning tokens within /review or sometimes review will just stop mid-way due to some policy violation out of no where.

>> Code review started: current changes <<

• Explored
  └ Read SKILL.md, AgendaDayBucketing.swift, AgendaDayBucketingTests.swift,
    Search addingDays\(|startOfDay\( in Model
    Search @interface BCCalendarDate|class BCCalendarDate in SharedCode
    Read DateUtilities.h, agenda_tests.log
    Search (Test Suite|Test Case|BUILD SUCCEEDED|BUILD FAILED|** TEST) in agenda_tests.log
    Search (Test Suite|Test Case|BUILD SUCCEEDED|BUILD FAILED|TEST SUITE|TEST FAILED|TEST SUCCEEDED) in agenda_tests.log

■ exceeded retry limit, last status: 429 Too Many Requests, request id: 501748cb-65ab-43a7-9dae-cde1e55b7a66

What steps can reproduce the bug?

I have turned off sub-agents (I tried this a few days ago and codex became extremely unstable, crashing / getting lost between threads / some sub-agents just vanishing mid-way and orphaning identifiers and so on). Apologies I haven't had the opportunity to log those bugs but I'd like to not use sub-agents. In practice I haven't yet found these to provide additional value to my current workflows.

Any way - point being, review seems to still be using Search(...) and other tools which potentially utilize sub-agents. I'm running into 429 errors a lot.

After I did: /review current changes but do NOT use sub-agents, do everything yourself the review finally worked.

What is the expected behavior?

_No response_

Additional information

_No response_