Summary
Fahd Mirza provides a step-by-step demonstration of Google’s newly released Agents CLI, an open-source tool designed to dramatically simplify building, evaluating, and deploying production-grade AI agents on Google Cloud. The video covers the full setup process — including Node.js, UVX, gcloud CLI, and Gemini CLI prerequisites — and shows how a single install command injects seven skills into a terminal coding agent: scaffold, build, eval, deploy, observability, publish, and workflow.
The architectural walkthrough explains how Agents CLI orchestrates Google’s full agentic stack under the hood: LLM selection (Gemini, Claude, OpenAI, or Model Garden), agent orchestration via Google’s ADK and A2A protocol, evaluation pipelines through Agent Platform Evaluation, and deployment to Cloud Run or GKE for larger-scale workloads. Supporting infrastructure — OpenTelemetry observability, Terraform and GitHub Actions for CI/CD, and BigQuery for RAG analytics — is configured automatically without manual service setup.
As a live test, Mirza prompts Gemini CLI to build and deploy a customer support agent that handles error messages, installation issues, and basic debugging guidance, showing the tool handle authentication, scaffolding, and Cloud Run deployment from a single natural language instruction. The video is a practical entry point for engineers looking to move from local agent prototypes to fully deployed, observable cloud agents using Google’s infrastructure — and notes that the free tier covers most of the setup, with costs only beginning at actual Cloud Run deployment.
📺 Source: Fahd Mirza · Published April 27, 2026
🏷️ Format: Hands On Build
