LEO's skills · 6 active · streams metrics via Datadog MCP ⌘
full library →Test Configuration
ready
0%
Peak RPS
0
Avg p95
0ms
Peak p99
0ms
Error rate
0.00%
Throughput · RPS
0
Latency · p95
0
Virtual Users
0
Errors / sec
0
How LEO works
01
Generate realistic load
LEO models user behavior — not just request floods. Sessions, think time, payload variance, geo distribution.
02
Watch the right signals
RPS and error rate are table stakes. LEO correlates p99 deviations with infrastructure metrics (pool saturation, GC pauses) to find the actual bottleneck.
03
Report to ARIA
Performance findings are tagged and forwarded — capacity ceilings, regressions, and scaling thresholds become part of the unified quality story.