Looker CA API with Google ADK Streaming

This skill provides guidance for integrating the Looker Conversational Analytics (CA) API with the Google Agent Development Kit (ADK) to build sophisticated, streaming data agents.

Overview

By combining the CA API's natural language-to-query capabilities with ADK's orchestration and streaming features, you can create applications that:

• Stream Real-time Responses: Provide immediate feedback as the CA API processes and generates data.
• Orchestrate Multi-step Workflows: Chain data fetching with analysis, visualization, or external tool calls.
• Implement Conditional Logic: Use run_when predicates to execute sub-agents only when specific data conditions are met.
• Normalize Session State: Share data between agents using ADK's session.state.

1. Prerequisites

• Looker Instance: Configured with the CA API enabled and appropriate IAM roles.
• Google ADK: Installed (pip install google-adk).
• GCP Project: Vertex AI and Gemini Data Analytics APIs enabled.
• Environment Variables: LOOKERSDK_BASE_URL, LOOKERSDK_CLIENT_ID, LOOKERSDK_CLIENT_SECRET, GCP_PROJECT_ID, GCP_LOCATION.

2. Implementation Pattern: CA Query Agent

The ConversationalAnalyticsQueryAgent bridges the CA API with ADK's streaming interface.