openai / openai/codex

Daybreak task state is not exposed when launching delegated reviewers

Open
#46,605 2 comments 1 reaction 0 assignees View on GitHub

Nobody has claimed this yet.

app bug subagent
Dominant language
Rust
Stars
125k
Forks
19.4k
PR merge metrics
PR metrics pending

Description

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

Version 26.915.31945

What subscription do you have?

Pro 20x

What platform is your computer?

macOS 26.6.2 (Build 25G83), Apple silicon (arm64)

What issue are you seeing?

Daybreak is enabled using the visible switch in the task header, but delegated task launchers do not expose the Daybreak state or provide a Daybreak parameter.

Previously, I enabled Daybreak at the task level and successfully launched a reviewer using gpt-5.6-sol with High reasoning while the Daybreak switch was visibly enabled.

Now an agent cannot determine whether Daybreak is active or propagated to a delegated reviewer. The only programmatically visible configuration may report desktop.daybreak-enabled = false even when the task-level switch is enabled, so it is not an authoritative signal.

Launching gpt-5.6-sol successfully also does not prove that the delegated reviewer is running with Daybreak. This makes the feature unverifiable and prevents reliable agent orchestration with Daybreak.

What steps can reproduce the bug?
  1. Open a task in the Codex App.
  2. Enable Daybreak using the switch in the task header.
  3. Ask the task to launch a delegated reviewer using gpt-5.6-sol with High reasoning.
  4. Inspect the available delegated-task or subagent launch parameters.
  5. Observe that there is no Daybreak parameter or readable task-level Daybreak state.
  6. Launch the reviewer.
  7. Observe that the result only identifies the base model and reasoning level, without confirming whether Daybreak was applied or inherited.

Attempting to use gpt-daybreak-blue-latest directly is not an equivalent workaround. It can fail with an access_programs.cyber=daybreak_blue entitlement error because Daybreak is a task mode combined with a base model, not simply that model identifier.

What is the expected behavior?

When Daybreak is enabled for a task, delegated reviewers should either inherit that mode automatically or allow the caller to request it explicitly.

The Codex App should expose the effective Daybreak state for the delegated reviewer so the calling task can confirm that Daybreak was applied. A successful launch using gpt-5.6-sol should not be the only available signal.

Additional information

The visible switch in the task header appears to be the authoritative Daybreak control. The global desktop.daybreak-enabled configuration value can conflict with the visible task-level state.

In the earlier successful workflow, the reviewer launched while the Daybreak switch was visibly enabled. However, the delegated reviewer did not expose independent confirmation that it had inherited Daybreak.

01a0b853-1f2f-7990-997e-b101eb7c2e57

Contributor guide

Open the contributing guide

First steps

  1. Read the whole issue, then the project's contributing guide.
  2. Comment on the issue to say you are picking it up — it saves two people doing the same work.
  3. Fork the repository and make your change on a branch.
  4. Open a pull request that references the issue number.

Research direction

Start by tracing the delegated-task or subagent launch parameters from the task header's Daybreak switch, then compare their effective state with desktop.daybreak-enabled. Done means a delegated reviewer either inherits or accepts Daybreak explicitly and exposes confirmation of the effective mode, rather than only its base model and reasoning level.

Written by the indexing model from the issue text.

Assessment

Tech stack
rust
Domain
ai, devtools
Issue type
Bug
Difficulty
4/5
Estimated time
3-5 days
Activity status
Active
Clarity
Mostly clear
Newbie friendliness
45/100

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.