Skip to content
Kitbay

dd-instrument-llmo

MITUpdated on Nov 9, 2026185GitHub stars

A Datadog skill that adds LLM Observability to a Python or Node.js/Next.js backend that calls language models or runs AI agents.

What the skill teaches

A Datadog skill that adds LLM Observability to a Python or Node.js/Next.js backend that calls language models or runs AI agents. The agent first analyses the project: the runtime, any LLM framework, the web framework, existing instrumentation and how session IDs flow, and which dependency file the deployment uses. It then sets up credentials, initialises ddtrace or dd-trace with the right options, adds agent and workflow spans and links browser RUM sessions to LLM traces, records the dependency, fixes gaps in session ID handling and reports how to verify the result.

Capabilities

  • Detect the LLM runtime and framework in a project
  • Initialise ddtrace or dd-trace for LLM Observability
  • Add agent and workflow spans
  • Find and fix gaps in session ID propagation