Trace and debug agent runs with AgentOps
Instrument agent sessions so operators can replay runs, track model costs, inspect tool calls, and compare behavior across frameworks.
npx skills add agentskillexchange/skills --skill trace-and-debug-agent-runs-with-agentops
Use AgentOps when an agent workflow needs run-level observability before it is trusted in production or repeated evaluations. The operator installs the SDK, initializes tracing at the start of the agent process, captures LLM calls and tool activity, and reviews session replays, costs, and failure patterns after the run. Invoke this instead of using the AgentOps product normally when the work is adding repeatable observability to an agent pipeline, not browsing dashboards by hand. The scope boundary is agent-session tracing, replay, cost monitoring, and debugging instrumentation across supported runtimes; it is not a generic analytics platform or a replacement for broader infrastructure monitoring.
What this skill actually does
Inputs and prerequisites: AgentOps Python SDK or supported framework integration, AgentOps API key or self-hosted deployment, agent runtime using OpenAI Agents SDK, CrewAI, Agno, LangChain, AutoGen/AG2, CamelAI, or another supported integration.
Setup notes: Install the AgentOps SDK with pip install agentops, configure an AgentOps API key or self-hosted endpoint, initialize AgentOps at process startup, end sessions explicitly, and review replay, cost, and tool-call traces after representative agent runs.
Source and verification boundary: use https://docs.agentops.ai as the canonical reference before running the workflow; keep commands, API calls, CLI usage, and generated outputs reviewable against that upstream source.
Framework fit: publish this as a Multi-Framework workflow only when the operator can invoke the documented toolchain directly, rather than treating the upstream project as a generic product listing.