Pi starts with one calm footer line: model, context use, tokens, cost, all from different extensions, laid out once.
You ask something; while the model answers, the segments move together: tokens climb, cost climbs, context fills.
A second question keeps accumulating; the footer is a running total, not a per-turn flash.
/footer hides the whole line when you want the screen; /footer again brings it back.
Protagonist: footer. Supporting cast: cost, context-size, model-info, tps.
The point is the shared event bus: one renderer, many producers, no layout fights.
# footer demo
What a human should take away in ten seconds:
1. Pi starts with one calm footer line: model, context use, tokens, cost, all from different extensions, laid out once.
2. You ask something; while the model answers, the segments move together: tokens climb, cost climbs, context fills.
3. A second question keeps accumulating; the footer is a running total, not a per-turn flash.
4. `/footer` hides the whole line when you want the screen; `/footer` again brings it back.
Protagonist: footer. Supporting cast: cost, context-size, model-info, tps.
The point is the shared event bus: one renderer, many producers, no layout fights.