codex - 💡(How to fix) Fix Codex Desktop: hidden context indicator still blocks long-session context management

What version of the Codex App are you using?

OpenAI.Codex 26.519.11010.0 on Windows x64.

Session metadata from local JSONL:

• originator: Codex Desktop
• cli_version: 0.133.0
• model_provider: openai
• active model in app: gpt-5.5

Local CLI on PATH reports:

• codex-cli 0.130.0-alpha.5

What issue are you seeing?

The Codex Desktop app no longer shows an always-visible context/token usage indicator after the update.

I found existing issue #23794, but that issue is locked, so I could not add my data point there: https://github.com/openai/codex/issues/23794

This is still a workflow problem. The data exists in local session logs, but the app no longer exposes it passively in the UI.

Latest local token_count event from my active Desktop session reports:

model_context_window = 258400
last_input_tokens ~= 150000
last_total_tokens ~= 150500

Why /status or /compact is not enough

The missing indicator was useful because it was passive context pressure, not just a command result.

In long coding sessions, I need to know when context pressure is growing so I can decide whether to compact, fork, or start a fresh thread before quality drops or compaction surprises me. A command I have to remember to run is not equivalent to a small always-visible signal near the composer/status area.

Expected behavior

Please restore an always-visible context indicator, or add an opt-in advanced setting such as:

Show context pressure in composer

The indicator could be approximate. It only needs to make context pressure visible, clearly separated from quota/rate-limit usage.

Suggested display examples:

• Context: 58% used
• Context: medium pressure
• Context: 150k / 258k

Additional context

I understand from #23794 that this removal may have been intentional and related to encouraging automatic compaction. The issue is that automatic compaction is not a complete replacement for visibility. Power users still need passive context awareness to plan long-running development work.