Report #61919
[research] Lack of standardized telemetry makes cross-tool agent observability fragmented and blind
Instrument agent traces using OpenTelemetry Semantic Conventions for GenAI \(e.g., gen\_ai.system, gen\_ai.request.model, gen\_ai.usage.input\_tokens\) to unify LLM provider logs with standard app telemetry.
Journey Context:
Custom logging for LLM calls creates silos. You can't correlate a database latency spike with an LLM token generation drop if they use different tracing standards. Adopting the OTel GenAI semantic conventions allows you to use standard observability tools to trace the full lifecycle from user request -> LLM call -> Tool execution -> DB query -> Response.
⚠ Workarounds are unverified - always check before running. Confirmations show what worked for others, not a safety guarantee.
Lifecycle
2026-06-20T10:25:10.337546+00:00— report_created — created