You can now investigate performance test results with the mabl MCP. Instead of comparing charts on the results page yourself, ask why a performance run passed or failed and get an answer grounded in the run's failure criteria and load metrics.
- Check each failure criterion: see the threshold, the observed value, and the result for every workload.
- Find the slow or failing endpoint: break API workloads down by endpoint, with response time percentiles and responses by status code.
- Find the slow or failing browser step: break browser workloads down step by step, with step durations, web vitals, and the network requests each step made.
- Catch a misleading pass: a performance test with no failure criteria always passes, and a test can pass its criteria while functional iterations fail. The results call out both cases so they don't read as healthy.
Try it out
Performance results are available through the mabl cloud MCP server with no setup beyond connecting your AI client. They include a live snapshot while a run is still in progress. Ask in plain language, for example:
"Why did the latest run of my checkout performance test fail? Which failure criterion was breached?"
"Which API endpoint was slowest at the 95th percentile in this performance run, and what status codes did it return?"
"Which browser step is slowing down this performance test?"
Conversational results analysis also supports performance test runs through the mabl MCP. Ask the mabl agent why a run passed or failed, and it cites the failure criteria and metrics behind the result.