What observability tools does this skill support?

This skill supports Google Cloud Trace, BigQuery Agent Analytics, AgentOps, Phoenix (Arize), and Weave (W&B).

Does this skill help track LLM costs?

Yes, integrations like AgentOps and BigQuery Analytics are designed to capture token counts and metadata that allow for detailed cost and performance analysis.

Can I log multimodal content like images and files?

Yes, the BigQuery patterns include configurations for offloading large multimodal content to Google Cloud Storage while maintaining links in your analytics tables.

Is this skill compatible with local agent development?

Absolutely. Scripts are provided for local setup, provided you have the necessary Google Cloud project credentials or third-party API keys.

How do I verify if my observability setup is working?

The skill includes a 'validate-observability.sh' script that checks environment variables, connectivity, and IAM permissions for all configured tools.

Google ADK Observability Patterns

Name: Google ADK Observability Patterns
Author: vanman2024

byvanman2024

分析と監視

Implements comprehensive monitoring, logging, and tracing patterns for Google ADK agents across cloud-native and third-party platforms.

概要

This skill provides production-ready templates and automated scripts for integrating deep observability into Google ADK-based AI agents. It streamlines the setup of Google Cloud-native services like Cloud Trace and BigQuery Agent Analytics, while also supporting popular third-party tools such as AgentOps, Phoenix, and Weave. Whether debugging agent logic, tracking LLM costs, or monitoring performance at scale, this skill offers the necessary configurations, schemas, and validation tools to ensure robust agent monitoring and operational excellence.

主な機能

0 GitHub stars
Native integration templates for AgentOps, Phoenix, and Weave
Built-in validation scripts for checking connectivity and IAM permissions
Support for multimodal content logging with GCS offloading
Automated setup scripts for Cloud Trace and BigQuery Analytics
Pre-defined BigQuery schemas for comprehensive agent event logging

ユースケース

Debugging complex agent reasoning and tool-calling behavior
Building comprehensive audit trails for agent-user interactions via BigQuery
Monitoring production LLM performance, latency, and token costs

概要

主な機能

0 GitHub stars
Native integration templates for AgentOps, Phoenix, and Weave
Built-in validation scripts for checking connectivity and IAM permissions
Support for multimodal content logging with GCS offloading
Automated setup scripts for Cloud Trace and BigQuery Analytics
Pre-defined BigQuery schemas for comprehensive agent event logging

ユースケース

Debugging complex agent reasoning and tool-calling behavior
Building comprehensive audit trails for agent-user interactions via BigQuery
Monitoring production LLM performance, latency, and token costs