Inconsistent project visibility and fragmented context between ChatGPT and Codex

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

26.901.41123

What subscription do you have?

Pro

What platform is your computer?

macOS

What issue are you seeing?

Inconsistent project visibility and fragmented context between ChatGPT and Codex

Switching between ChatGPT and Codex makes it difficult to understand where my projects, conversations, and skills belong.
A concrete example from the macOS desktop app:

• Horizon appears in both interfaces, while HPC appears in ChatGPT but is missing from the Codex Projects sidebar.
• The Codex sidebar shows four projects, all identified as local projects by the app’s project-list tool. HPC is identified as a ChatGPT project.
• A Codex conversation associated with my ChatGPT project “Weekly” remains accessible under Recents, even though Weekly itself is missing from the Codex Projects sidebar.
  The assistant’s project-list tool returns both project types regardless of the selected interface. This initially caused the assistant to incorrectly suggest that HPC was visible, until I provided a screenshot.
  Whether intentional or a bug, this behavior is confusing: a conversation can be accessible in Codex while its parent project is absent from project navigation. The UI does not explain why.
  Skills add another layer of uncertainty: it is unclear which skills are shared, which are local, and which are actually available in the current conversation.
  Please consider:

1. Consistent project navigation across ChatGPT and Codex, with explicit labels for local and cloud-backed projects.
2. An explanation and a clear next action when a project is unavailable in an interface.
3. Visible indicators of the current conversation’s project, execution environment, and loaded context.
4. A unified skills view showing scope, availability, and dependencies.
5. Letting users continue one task from discussion through execution without losing project continuity.
   Different execution environments make sense. Having to reconstruct the same work across separate interfaces does not.

What steps can reproduce the bug?

Feedback ID: 01a07978-11c3-7930-9dd1-5ea1c7e69d3d

What is the expected behavior?

No response

Additional information

No response

Reproduce the behavior in the macOS desktop app using the reported Feedback ID 01a07978-11c3-7930-9dd1-5ea1c7e69d3d, comparing ChatGPT and Codex project sidebars, Recents, and skills visibility. No files or tests are named, and completion would require defining consistent navigation, context indicators, and skills scope across both interfaces.