Desktop: add an “always expanded” option for progress and reasoning activity

Product and environment

• ChatGPT/Codex desktop app: 26.715.52143
• Platform: macOS
• Model currently used: GPT-5.6 Sol
• Desktop conversation detail mode: STEPS_COMMANDS

Feedback

Recent desktop updates present ongoing Codex work under fewer, collapsed action headers. A few days ago, the activity stream exposed substantially more of the model-provided progress commentary and reasoning summaries while a task was running. The newer grouped presentation makes it harder to follow what Codex is checking, what it has learned, and what it plans to do next.

Please add a user-facing conversation-display preference such as:

• Always expand progress/activity
• Show model-provided progress commentary and reasoning summaries as they arrive
• Keep command and tool details expanded while a task is running
• Preserve the live chronological stream instead of consolidating it into a few generic headers
• Make this display preference independent of model reasoning effort

The existing conversationDetailMode = "STEPS_COMMANDS" setting does not restore the earlier, more continuously visible presentation. Prompting every task to provide frequent progress updates helps the content but does not solve the collapsed UI.

This is not a request to reveal private chain-of-thought. It is a request to keep the progress commentary, reasoning summaries, commands, and tool activity that Codex already chooses to expose visible and optionally expanded.

Why this matters

For long-running engineering tasks, the live activity stream is part of the product's trust and usability: it lets users notice a wrong assumption, redirect work early, and distinguish useful progress from a stalled task. An “always expanded” preference would support users who value that observability while preserving the compact default for everyone else.