Monitor your AI agents with powerful observability tools. Observability helps you see what your agent does, how fast it runs, and where errors happen. This guide covers the main options for tracking agent activity.
There are four main observability tiers. Cloud Trace shows the full execution flow with distributed tracing. Prompt-Response Logging captures GenAI interactions and stores them in Google Cloud. BigQuery Agent Analytics gives structured data for custom dashboards. Third-Party Integrations let you use tools like AgentOps or Phoenix.
Use this guide to set up monitoring for debugging, performance improvement, and compliance. It works for deployed agents and local development.
Global
mkdir -p ~/.claude/skills/adk-observability-guideProject
mkdir -p .claude/skills/adk-observability-guideSource Repository
Appinsights Instrumentationmicrosoft/azure-skills
Learn how to add Azure Application Insights monitoring to your web app
Azure Kustomicrosoft/azure-skills
Run fast KQL queries on Azure Data Explorer for log and telemetry analysis
Caveman Statsjuliusbrussee/caveman
See real token usage and estimated savings directly from your session log
Azure Observabilitymicrosoft/azure-skills
Track Azure app errors, performance, and infrastructure with monitoring tools
Convex Performance Auditget-convex/agent-skills
Find and fix Convex performance problems with step-by-step diagnostic guidance
Google Agents Cli Observabilitygoogle/agents-cli
Set up tracing, logging, and analytics for your ADK agents easily
Golang Observabilitysamber/cc-skills-golang
Monitor your Go services with logs metrics traces and profiling
Golang Samber Slogsamber/cc-skills-golang
Compose, sample, and route Go logs to multiple backends with ease