The HTTP Metrics tab shows how the HTTP calls in your flows are performing, so you can find a failing or slow integration without opening individual conversations. It counts only calls made by the HTTP node. API calls made by an AI agent's tools are reported under AI Agent Metrics.
The tab appears only when your deployment has Prometheus. If you expect to see it and it is missing, ask your administrator.
| Widget | What it shows |
|---|---|
| Total Requests | The number of HTTP node requests in the range. |
| Total Errors | The number of requests that returned an error status code. |
| Error Rate | Failed requests as a share of all requests. |
| Average Duration | The average response time. |
| Request Timeline | Total and failed requests over time. |
| Status Code Breakdown | Each status class's share of responses: 1xx, 2xx, 3xx, 4xx, 5xx and Other. Hover over a slice to see the individual codes. |
| Response Duration (ms) | Average, 95th percentile and 99th percentile duration over time. |
| Request Type Breakdown | Synchronous against asynchronous requests, depending on how each HTTP node is set to execute. |
| Metrics by Endpoint | A table with one row per endpoint and method: Method, Endpoint, Total, Errors, Error % and Avg. Duration. Each endpoint appears as the URL configured on the node, so calls that differ only by a variable value share one row. |