Phase 6: Display Binding Windows Per Account — Don't Let Aggregates Hide Exhausted Keys #41

New Issue

Closed

opened 2026-05-21 00:49:48 -05:00 by null · 0 comments

null commented

2026-05-21 00:49:48 -05:00

Owner

References: remaining-usage-accuracy-review-plan.md — Phase 6

Summary

The frontend aggregates token limits by summing gateway limits and chooses the soonest reset. This is only accurate if every gateway/account shares the same provider window and limit semantics. A nearly exhausted provider account can be hidden by another account's available capacity.

Acceptance Criteria

A nearly exhausted Claude account is visible even if another gateway/account has room
Aggregate numbers are clearly labeled and do not replace binding per-account windows

Work

Show provider/account windows independently before any aggregate
Aggregate only same-unit, same-window values
If windows differ, show the most constrained account/model as the primary remaining usage indicator
Keep per-account details visible alongside any aggregate summary

Do Not

Do not sum unrelated windows just to produce one global percent
Do not let a healthy account mask an exhausted account
Do not choose only the soonest reset if the limit units or providers differ
Do not remove per-account details after adding an aggregate summary

Priority: Medium
Phase: 6

References: [`remaining-usage-accuracy-review-plan.md`](docs/remaining-usage-accuracy-review-plan.md) — Phase 6 ## Summary The frontend aggregates token limits by summing gateway limits and chooses the soonest reset. This is only accurate if every gateway/account shares the same provider window and limit semantics. A nearly exhausted provider account can be hidden by another account's available capacity. ## Acceptance Criteria - A nearly exhausted Claude account is visible even if another gateway/account has room - Aggregate numbers are clearly labeled and do not replace binding per-account windows ## Work 1. Show provider/account windows independently before any aggregate 2. Aggregate only same-unit, same-window values 3. If windows differ, show the most constrained account/model as the primary remaining usage indicator 4. Keep per-account details visible alongside any aggregate summary ## Do Not - Do not sum unrelated windows just to produce one global percent - Do not let a healthy account mask an exhausted account - Do not choose only the soonest reset if the limit units or providers differ - Do not remove per-account details after adding an aggregate summary **Priority:** Medium **Phase:** 6