google/agents-cli

What it solves

It simplifies the process of building, scaling, governing, and optimizing enterprise-grade agents on the Gemini Enterprise Agent Platform. It removes the need for developers to manually learn every specific CLI and service required to deploy agents on Google Cloud by providing a standardized set of tools and "skills" that can be used by both humans and AI coding assistants.

How it works

agents-cli provides a command-line interface and a set of specialized skills that integrate with coding agents (like Claude Code or Antigravity CLI). These skills teach the coding agent how to handle the development lifecycle, use the ADK Python API for orchestration, scaffold projects, run evaluations, and manage deployments to Google Cloud services like Cloud Run or GKE.

Who it’s for

Developers building AI agents using the ADK framework who want to automate the development-to-deployment pipeline on Google Cloud, as well as users of AI coding assistants who want their assistants to be experts in Google's agent platform.

Highlights

• Coding Agent Integration: Specifically designed to give AI coding assistants the skills to build and deploy agents.
• Evaluate and Optimize: Includes tools to run inference over datasets, grade traces using LLM-as-judge, and auto-tune prompts based on evaluation data.
• End-to-End Lifecycle: Covers everything from project scaffolding (create) and local development (run) to cloud deployment (deploy) and registration with Gemini Enterprise (publish).
• Flexible Infrastructure: Supports deployment to Agent Runtime, Cloud Run, and GKE, with built-in support for CI/CD pipelines.