bigquery-agent-analytics-sdk

You can analyze AI agent traces in BigQuery by initializing the Python client to connect to your Google Cloud project, generating observability dashboards that visualize agent performance metrics and decision trajectories at scale.

LLM-as-Judge evaluation assesses AI agent performance by using a language model to grade agent outputs. This Skill applies LLM-as-Judge evaluation to BigQuery traces to measure quality and detect behavioral regressions.

Yes, running BigQuery agent analytics requires Python 3.10 or higher. You also need a Google Cloud project with BigQuery enabled and the google-cloud-bigquery dependency installed to process agent traces.

Yes, you can extract AI agent decision traces using the Agent Context Graph feature. It queries BigQuery to map and extract complex decision trajectories for observability and drift detection.

The best way to detect agent behavior drift at scale is by analyzing historical traces in BigQuery. This Skill provides trajectory matching and advanced analytics to identify behavioral regressions across production agents.

To retrieve a single AI agent trace from BigQuery, initialize the Python analytics client and call the get_trace function with the specific trace ID, such as client.get_trace("trace-abc-123").