Build an AI Agent and Configure an MCP Server on Cloud Run

Why it matters

Cloud Run is a flexible application runtime, but that flexibility leaves more packaging, scaling, and integration choices with the team than a purpose-built managed agent runtime.

What to practise

Build an agent workflow, prepare it for container deployment, configure its MCP integration, deploy with the ADK CLI, and test the running service.

How it fits

Compare the result with the Agent Runtime and GKE options in subsection 4.2. Focus on the control you gained, the operations you inherited, and the workload constraints that justify both.