Google Ships agents-cli to Turn Any Coding Assistant Into an ADK Expert
Google's agents-cli injects 7 expert skills into any coding assistant, turning it into a full-stack ADK agent engineer that can scaffold, evaluate, and deploy to Google Cloud
Takeaways
- Google released agents-cli, a CLI + 7-skill pack that turns coding assistants (Claude Code, Gemini CLI, Codex) into ADK deployment experts on Google Cloud.
- Skills inject structured expert knowledge into the coding assistant at runtime, covering scaffolding, ADK code patterns, evaluation, deployment, and observability.
- Works with any coding agent supporting the Agent Skills spec; also runs fully standalone from the terminal without a coding assistant.
- Local development requires only a free AI Studio API key; Google Cloud is only needed for deployment to Agent Runtime, Cloud Run, or GKE.
- Limitations: tightly coupled to ADK and Google Cloud; generated code still needs human review before deploying, especially eval criteria and security settings.
- Released at Google Cloud Next under Apache 2.0, currently v0.1.3 (pre-GA); available via PyPI with one-command install.
Building an AI agent is one thing. Knowing which of the 40+ components in Google's Agent Platform to use, in what order, with what configuration -- that's the part that actually takes time. agents-cli is Google's answer to that problem: a CLI and a set of seven specialized skills that you install into your coding assistant, turning it into an expert at the full ADK agent lifecycle from first scaffold to production deployment.
A tool that teaches tools
agents-cli provides a unified, machine-readable interface for AI development tools to interact with ADK, encapsulating expert knowledge of ADK, agent evaluation, and deployment to Google Cloud so that coding assistants can perform complex actions from natural language prompts. The key insight is that this is not another agent framework, not another SDK -- it is a tool that teaches tools.
The hard part about building an agent in 2026 is not writing SDK code -- it is knowing which of the 40+ Agent Platform components to use, in what order, with what configuration. Scaffolding a project, wiring up evals, configuring deployment targets, registering with Agent Platform -- each step has its own docs, its own CLI commands, its own gotchas. agents-cli collapses all of that into a set of skills your coding agent can read and execute.
The skills layer
agents-cli is a CLI and skills package for building, evaluating, and deploying AI agents on Google Cloud. Agents are built with Google's Agent Development Kit (ADK) -- agents-cli handles everything around it: scaffolding, evaluation, deployment, and observability. ADK (Agent Development Kit) is Google's open-source framework for defining agent logic, tools, and orchestration in Python. agents-cli is the layer that wraps ADK and knows how to wire it into the rest of Google Cloud.
agents-cli bundles 7 skills that give your coding agent deep knowledge across the full ADK lifecycle. Here's what each one covers:
- google-agents-cli-workflow -- Development lifecycle, code preservation rules, model selection guidance
- google-agents-cli-adk-code