Quota intelligence
Understand what your quota bought.
This renderer reads only cqa-report-v1. It does not know how quota, Guardian, banked-reset, or workflow statistics were calculated.
Quota efficiency
Observed work per quota point · regime context remains attached to every cohort
Click a cohort to inspect interval, evidence, composition and regime.
What deserves attention
Deterministic findings from the report.
Policy history & regimes
Detected model-specific regime windows. Select a regime to focus the quota-efficiency view.
Select a regime window to filter quota cohorts; click again or clear the filter to restore all cohorts.
Guardian / Approve for me
Typical periods, tail periods and observational uncertainty.
Click a point to inspect the period and its largest paired approval episodes.
Banked reset audit
Whole-period capacity and equal-quota boundary behavior.
Click a slice to inspect its ratio and compare all slice sizes.
Workflow performance
How models responded, how quickly work moved, and where the workload went.
Model performance
Compare response efficiency without charging models for external tool and wait time. Reasoning and response latency remain included; visible generation is shown separately.
evidence-aware
Response timing
How quickly did model responses start and tasks finish?
Work by role
How token work was distributed across explicitly attributed roles.
Workflow summary
Observational signals that can guide investigation; they are not waste labels.
Click a signal to inspect its definition and evidence.
Workflow timeline
See when agents were active, when compactions happened, and where pauses or overlaps occurred.
■Guardian activity interval│Compaction point event▤2+ child overlap interval▧Quiet/pause interval◆Handoff point event
Point events use horizontal position only; marker width is not duration. Intervals use segment width to show elapsed wall time; very short intervals receive a minimum visible width.
Agent activity
Inspect agent lifetimes, compactions, concurrency, and directly attributed usage.
Highest-cost agents
Based on directly attributable priced usage.
Recent compactions
Data quality
What CQA could measure reliably, and where evidence is incomplete.
| Cohort | Regime | Quota points | Episodes | Model consistency | Reasoning-effort consistency | Status |
|---|