Agent Observability Setup
$2.99OfficialImplement tracing, logging, and monitoring for AI agent systems: LLM call traces, tool execution logs, latency tracking, and replay debugging.
Model requirements
advanced
33k tokens
Preview
When to use
Use this skill when the user needs visibility into what an AI agent is doing: "I can't tell why my agent made that decision," "add tracing," "debug a failed agent run," "set up monitoring for our agent system." Trigger phrases: observability, tracing, agent traces, LLM monitoring, replay debugging, agent logs, LangSmith, OpenTelemetry.
Do NOT use it for application logging in general (use standard APM), or for evals (related but distinct — observability is recording; evals are measuring). Use this when the specific challenge is the opacity of multi-step, tool-calling agent behavior.
Inputs to gather
- The agent system: framework (LangGraph, CrewAI, raw function-cal
…
🔒 Buy this skill once ($2.99) to unlock the full playbook and install it anywhere — or get everything with Pro.