Visual Indicator for Codex CLI

[maheshrijal]

Currently, I have no easy way to metrics without breaking my workflow.

I'd like to suggest adding a clean, unobtrusive visual indicator in the terminal UI that shows:

1. Context window utilization:
   - A simple progress bar showing how full my context window is
   - Color shifts as I approach capacity
2. Token and cost awareness:
   - Current token count for the session
   - Estimated cost based on my model
   - Perhaps a small indicator when I'm approaching limits
3. At-a-glance performance insights:
   - Something to help me understand if my context size is impacting response times

[tibo-openai]

Would like to start with 1 & 2. I believe these are either in progress or there are PRs that have been submitted already for those and we're close to have a serious improvement here, could be useful to track all the in progress work and summarize it here to make sure nothing falls through the cracks.

For 3, the context size will inevitably impact response times, but I'm not sure there is a good way to represent this or whether it is that useful.

[libertyteeth]

For 3, I think if 2 had "performance metrics gathering"[1] added to it, in addition to token and cost awareness, it would help to generate the reports for 3.

Two reports I can think of:

1. How much will my next query cost?
2. How long will my next query take?

Those 'reports' could be part of a "status line dashboard," which I would have configurable (multi-line; etc). So something like the following could be part of it:

... | NextQ $5.50/2m40s | ...

Perhaps a more detailed one, still tightly compressed space-wise; also, should be separately optionally configurable:

... | NextQ $5.50±0.13/2m40s±20s | ...

Thanks for listening!

[1] -- Metrics would be stored locally; I'd also have a configuration option for users to send sanitized metrics upstream, to help metrics for everybody improve, over time.