Trace LLM and agent workflows with OpenLLMetry
Add OpenTelemetry-based instrumentation to LLM and agent code so operators can inspect prompts, tool calls, latency, errors, and provider behavior in their existing observability stack.
npx skills add agentskillexchange/skills --skill trace-llm-and-agent-workflows-with-openllmetry
Use OpenLLMetry when an agent application needs vendor-neutral traces instead of another black-box dashboard. The workflow is to install the Traceloop SDK or the relevant OpenTelemetry instrumentations, initialize tracing in the application, send spans to the existing collector or observability backend, and inspect prompts, tool calls, latency, failures, and provider behavior during debugging or release review. The scope is instrumentation and trace export for LLM and agent workflows, not a generic OpenTelemetry or SaaS observability listing.
What this skill actually does
Inputs and prerequisites: Python, OpenLLMetry or Traceloop SDK, OpenTelemetry-compatible collector or backend.
Setup notes: Install the SDK with pip install traceloop-sdk, initialize tracing with the Traceloop SDK in the application, then route the emitted OpenTelemetry data to the team’s collector or observability backend.
Source and verification boundary: use https://traceloop.com/docs/openllmetry/getting-started-python 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.