Skip to content
Sign in
  • Browse all toolsFilter by category and pricing
  • Browse categoriesWriting, code, image, audio and more
  • Compare toolsPut up to three side by side
  • Read the blogGuides and comparisons

agents-cli

Open-source CLI and skills for building, evaluating, and deploying ADK agents on Google Cloud.

agents-cli is a command-line tool and skills collection that scaffolds, evaluates, and deploys ADK agents on Google Cloud. Designed for developers using coding assistants or a terminal, agents-cli is available as an open-source GitHub project under the Apache-2.0 license, with installation through uvx or skills installation through npx.

Updated 9 Nov 20266 sections

agents-cli website screenshot

What is agents-cli?

agents-cli is an open-source CLI and skills collection for building, evaluating, and deploying AI agents with Google's Agent Development Kit. Published in the google GitHub organization under the Apache-2.0 license, agents-cli supplies commands and development guidance to coding assistants. Developers can also use the commands directly from a terminal without a coding agent.

agents-cli complements Antigravity CLI, Claude Code, Codex, and other coding agents rather than replacing them. ADK provides the agent framework, while agents-cli supplies the surrounding workflow for scaffolding projects, running evaluations, preparing deployment, and registering agents with Gemini.

What agents-cli can do

agents-cli covers the agent development lifecycle through dedicated commands and skills. The skills teach coding assistants ADK Python patterns, orchestration, callbacks, state, evaluation methodology, deployment, and observability. Evaluation support includes metrics, datasets, LLM-as-judge methods, and adaptive rubrics, alongside commands that operate on agent traces and evaluation results.

  • agents-cli creates new agent projects and upgrades existing projects.
  • agents-cli generates agent traces and grades them against metrics.
  • agents-cli synthesizes multi-turn evaluation scenarios for local agents.
  • agents-cli compares evaluation results and clusters failure modes.
  • agents-cli tunes agent prompts using evaluation data.
  • agents-cli deploys agents to Google Cloud and registers them with Gemini.

How agents-cli works

agents-cli requires Python 3.11+, uv, and Node.js. The documented setup command is uvx google-agents-cli setup; developers can alternatively install only the skills with npx skills add google/agents-cli. After installation, developers open their preferred coding agent and ask the assistant to build an agent using agents-cli.

agents-cli also supports a direct terminal workflow for scaffolding, evaluation, deployment, and other commands. Authentication uses Google Cloud or AI Studio. Local create, run, and eval workflows can use an AI Studio API key, while deployment skills cover Agent Runtime in Gemini platform, Cloud Run, GKE, CI/CD, and secrets.

Who agents-cli is for

agents-cli serves developers building ADK agents, including developers who want a coding assistant to handle the supporting commands and service workflows. Existing agent projects are also supported: scaffold enhance adds deployment or CI/CD configuration without requiring developers to start with a newly scaffolded project.

  • Developers can scaffold an ADK agent and test prompts locally.
  • Teams can evaluate agent behavior and investigate recurring failures.
  • Developers can add deployment and CI/CD to existing agent projects.
  • Teams can provision staging and production infrastructure through the CI/CD command.

agents-cli pricing and access

agents-cli is distributed as an open-source project under the Apache-2.0 license, with source code on GitHub and installation instructions using uvx or npx. The repository does not list commercial plans for agents-cli. Google Cloud deployment creates resources in the user's own project, and the user is responsible for those resources under the Google Cloud Service Terms.

agents-cli strengths and limitations

agents-cli combines project creation, evaluation, deployment, and publishing in a workflow that supports both coding assistants and standalone terminal use. Local development does not require Google Cloud, and existing projects can gain deployment support. Setup dependencies, cloud requirements, and default telemetry are practical considerations when deciding how to use agents-cli.

  • agents-cli requires Python 3.11+, uv, and Node.js.
  • agents-cli requires Google Cloud for deployment and cloud features.
  • agents-cli emits usage telemetry by default; AGENTS_CLI_TELEMETRY or DO_NOT_TRACK can disable telemetry.

agents-cli FAQ

What is agents-cli?

agents-cli is an open-source command-line tool and skills collection for building, evaluating, and deploying ADK agents on Google Cloud. agents-cli supports coding assistants such as Antigravity CLI, Claude Code, and Codex, and also works directly from a terminal.

Is agents-cli open source?

agents-cli is an open-source project available on GitHub under the Apache-2.0 license. The repository provides installation instructions through uvx and a separate skills installation through npx. Deployed Google Cloud resources belong to the user's project and remain the user's responsibility.

Does agents-cli require Google Cloud?

agents-cli does not require Google Cloud for local development with create, run, and eval commands; developers can use an AI Studio API key to run Gemini with ADK locally. Deployment and cloud features do require Google Cloud.

Can agents-cli work without a coding assistant?

agents-cli works standalone from a terminal, so a coding assistant is not required. Developers can run scaffolding, evaluation, deployment, and other commands directly. The accompanying skills help coding assistants perform those workflows on the developer's behalf.

Can agents-cli work with an existing agent project?

agents-cli supports existing agent projects through the scaffold enhance command, which adds deployment or CI/CD configuration. agents-cli also provides scaffold upgrade to upgrade a project to a newer agents-cli version.

Does agents-cli collect telemetry?

agents-cli emits a usage signal per command invocation by default, recording metadata rather than file contents, arguments, or other input. Log entries stay in the user's Google Cloud project, while request headers carry usage metadata to Google. Environment variables can disable telemetry.

From the community

No reviews yet

Be the first to review this tool. What you write helps the next person choose.

More tools like agents-cli

Google ADK Go

www.github.com

An open-source Go toolkit for building, evaluating, deploying, and orchestrating AI agents.

Free

CC Switch

www.github.com

A desktop manager for switching AI coding providers and organizing MCP, Skills, and Prompts.

Free

Claude Code Templates

www.github.com

Open-source CLI for configuring Claude Code with agents, commands, MCP integrations, hooks, and monitoring tools.

Free

OpenRig

www.github.com

Build persistent AI coding agent teams with shared context, roles and coordinated work.

Free