@mayurajs/observability-datadog

Datadog APM or LLM Observability for Mayura's traces (mayura/exporter-otlp), through Datadog's OTLP intake. No Datadog Agent is needed.

npm install mayura @mayurajs/observability-datadog

import { agentRunTraceSpans } from 'mayura/exporter-otlp';
import { datadogTraceExporter } from '@mayurajs/observability-datadog';

const datadog = datadogTraceExporter({ apiKey, site: 'datadoghq.eu', llmObservability: { mlApp: 'support-bot' }, serviceName: 'support-agent' });
await datadog.sink(agentRunTraceSpans(observer.inspect(handle.id)?.recent ?? []), { signal: AbortSignal.timeout(10_000) });

• site is your Datadog site, datadoghq.com (US1) by default.
• Without llmObservability, runs go to APM as traces with their model and tool calls. With it, Datadog reads the GenAI attributes on Mayura's spans and shows agent, LLM and tool spans with models and token usage; mlApp names the application (the service name by default). LLM Observability keeps only spans with GenAI attributes, so workflow spans stay in APM. It is not available on ddog-gov.com.
• Spans are metadata only, so traces carry no prompts or outputs.
• Traces only: Datadog's OTLP metric intake takes delta sums, and Mayura's metric exporter sends cumulative ones.
• The key is passed in; nothing is read from the environment. Every other option is the OTLP trace exporter's.

See the observability guide. Apache-2.0.