llama_index.agent.step.duration

Multi-agent LlamaIndex systems lack visibility into agent handoffs, tool execution timing, and workflow failures without proper instrumentation, leading to slow debugging and degraded production reliability.

LlamaIndex query response times degrade at P95/P99 percentiles due to slow LLM calls, inefficient retrieval, or tool execution bottlenecks without granular latency breakdown.

Agent tool calls introduce significant latency due to slow external APIs, inefficient implementations, or excessive tool invocations, degrading overall agent responsiveness.