google-agents-cli-observability

The `google-agents-cli-observability` skill provides comprehensive guidance for monitoring and adding observability to deployed ADK (Agent Development Kit) agents. It is designed to help users set up tracing, configure logging, debug production traffic, and gain insights into agent behavior without requiring them to manually handle infrastructure complexities. This skill ensures that users can effectively monitor deployed agents, capture telemetry, and analyze agent interactions, thereby improving operational reliability and transparency in production environments.

The skill supports several key observability features including Cloud Trace for distributed tracing, prompt-response logging to capture GenAI interactions, and BigQuery Agent Analytics for structured event tracking. Cloud Trace works immediately without additional infrastructure, whereas logging and analytics require Terraform-provisioned resources such as service accounts, GCS buckets, and BigQuery datasets. Additionally, the skill provides guidance for integrating with third-party platforms like AgentOps, Phoenix, and MLflow, and includes reference files for setting up observability resources and verifying configurations. Clear instructions are provided for order-of-operations during deployments to avoid conflicts between SDK-deployed instances and Terraform-managed infrastructure.

Target users include developers and operators who manage ADK agents and need detailed monitoring and analytics capabilities. Typical use cases involve debugging latency issues, auditing LLM interactions for compliance, evaluating conversational analytics, and integrating agent telemetry into custom dashboards or external observability platforms. By following this skill, teams can ensure consistent observability across all agent deployments, streamline troubleshooting, and maintain compliance and performance oversight for production agents.