Performance

Call quality

The call quality card breaks down proxy performance on a per-agent basis. It surfaces success rates, latency percentiles, error distributions, and a paginated activity log so you can pinpoint exactly where quality is degrading.

Per-agent quality table

Each row represents one of your agents. The table columns show total calls, success percentage, and P95 latency. Cells are color-coded for at-a-glance assessment:

Metric	Green	Amber	Red
Success %	≥ 98%	90 – 97%	< 90%
P95 latency	≤ 500 ms	500 – 2000 ms	> 2000 ms

Expandable rows

Click any row to expand it. The expanded section shows a status breakdown — counts of 2xx, 4xx, and 5xx responses — and a mini bar chart visualizing the distribution.

Latency percentiles

The expanded row also displays three latency percentiles: P50 (median), P95, and P99. These values give you a complete picture of the latency distribution — the median shows typical performance while P99 reveals worst-case tail latency.

A large gap between P50 and P99 often indicates intermittent slowdowns caused by cold starts, rate limiting, or network congestion at the target endpoint.

Error breakdown

Below the latency percentiles, the error breakdown groups all non-2xx responses by HTTP status code. Each status code is shown with its count and a proportional bar representing its share of total errors. This makes it easy to tell whether failures are concentrated on one error type or spread across several.

Call activity log

At the bottom of the card, a paginated log lists individual proxy calls. Each entry shows the timestamp, caller agent, target agent, HTTP status, latency, and proxy type. Use the direction filter to show only incoming calls, only outgoing calls, or both.

The activity log is paginated server-side. Navigate pages using the controls below the list. Each page loads 25 entries.