AI Frameworks · google

agents-cli

agents-cli is a Google-backed CLI tool and skill suite that extends coding assistants (Claude, Antigravity, etc.) with commands to build, evaluate, and deploy AI agents on Google Cloud using the Agent Development Kit (ADK). It works standalone or integrated into your favorite coding assistant.

Talk to DEV.co Discuss Implementation

Source: GitHub — github.com/google/agents-cli

4.8k

GitHub stars

514

Forks

Python

Primary language

Apache-2.0

License (OSI-approved)

Key facts

Objective fields from the source. Values we can't verify are shown as “Unknown” rather than guessed.

Field	Value
Repository	google/agents-cli
Owner	google
Primary language	Python
License	Apache-2.0 — OSI-approved
Stars	4.8k
Forks	514
Open issues	26
Latest release	v1.0.0 (2026-07-01)
Last updated	2026-07-01
Source	https://github.com/google/agents-cli

What agents-cli is

Python 3.11+ CLI providing scaffolding, ADK code generation, evaluation frameworks (metrics, datasets, LLM-as-judge), deployment pipelines (Cloud Run, GKE, Agent Runtime), and observability integration (Cloud Trace, logging). Ships as PyPI package and npm skills; orchestrates Google Cloud APIs and CI/CD infrastructure.

Quickstart

Get the agents-cli source

Clone the repository and explore it locally.

terminalbash

1git clone https://github.com/google/agents-cli.git2cd agents-cli3# follow the project's README for install & configuration

Need it deployed, integrated, or customized instead? DEV.co ships production installs.

Best use cases

Rapid Agent Prototyping with Coding Assistants

Leverage Claude Code, Antigravity CLI, or other coding agents to scaffold and iterate on ADK agents without manual CLI learning. Skills encode patterns for agents, tools, orchestration, and callbacks.

End-to-End Evaluation and Optimization

Generate synthetic eval datasets, run agents over traces, grade with LLM-as-judge and custom metrics, analyze failure modes, and auto-tune prompts—all within a unified workflow.

Google Cloud Agent Deployment at Scale

Provision infrastructure (single-project or multi-stage CI/CD), deploy to Agent Runtime, Cloud Run, or GKE, manage secrets, and register with Gemini Enterprise Agent Platform from a single command set.

Implementation considerations

Requires Python 3.11+, uv package manager, and Node.js for skill integration into coding agents. Installation path varies: `uvx google-agents-cli setup` (full CLI) vs. `npx skills add google/agents-cli` (skills only).
Local development (scaffold, run, eval) works with AI Studio API key; cloud deployment requires active Google Cloud project with billing, IAM roles, and pre-configured secrets (DB credentials, API keys).
Eval workflow relies on dataset synthesis, multi-turn trace generation, and LLM-as-judge grading. Ensure eval datasets reflect production use cases; metric selection (custom rubrics vs. built-in) impacts grading quality.
CI/CD and staging/prod infrastructure provisioning (`agents-cli infra cicd`) creates multiple Google Cloud resources. Cost, permissions, and disaster recovery planning required before production rollout.
ADK patterns and observability integrations assume familiarity with agent callbacks, state management, and Cloud Trace/logging. Custom agents with non-standard patterns may not integrate cleanly.

When to avoid it — and what to weigh

Non-Google Cloud Deployments — agents-cli is tightly coupled to Google Cloud services (Cloud Run, GKE, Agent Runtime, Cloud Trace). If your agents must run on AWS, Azure, or on-prem, deployment and observability features are not applicable.
Agents Not Based on ADK — The tooling assumes ADK (adk.dev) as the agent framework. If you use LangChain, CrewAI, or custom frameworks, scaffolding, code generation, and skill patterns will not apply.
Teams Without Google Cloud Expertise — Despite automation, deployment, infrastructure provisioning (GKE, CI/CD pipelines), and observability still require Google Cloud IAM, networking, and service knowledge. Steep learning curve for unfamiliar teams.
Standalone Agents Without Coding Assistant Integration — While the CLI works standalone, the tool is optimized for use *through* a coding assistant. Manual CLI use case is possible but not the primary design intent; documentation emphasizes agent-driven workflows.

License & commercial use

Apache License 2.0 (Apache-2.0). Permissive OSI license allowing commercial use, modification, and distribution under the terms of the license (include NOTICE, state changes, provide license copy).

Apache 2.0 is a permissive OSI license that explicitly permits commercial use. You may build commercial products using agents-cli. However, Google Cloud service terms (Agent Runtime, Cloud Run, etc.) apply to deployed agents—review those separately. No per-seat licensing fees for the CLI itself.

DEV.co evaluation signals

Editorial assessment — not user reviews. Directional, with an explicit confidence level.

Signal	Assessment
Maintenance	Active
Documentation	Strong
License clarity	Clear
Deployment complexity	Moderate
DEV.co fit	Strong
Assessment confidence	High

Security considerations

agents-cli itself does not implement secret storage; it relies on Google Cloud Secret Manager for credential injection. Code generation for agents (scaffolding, ADK patterns) is template-based; no LLM-as-code-generator. Eval dataset synthesis and grading run within your Google Cloud environment. Ensure IAM policies restrict who can scaffold, deploy, and access agent traces. No explicit mention of input sanitization or SBOM; verify with maintainers for security audit requirements.

Alternatives to consider

Agent Development Kit (ADK) directly + manual Google Cloud CLI

ADK is the underlying framework; agents-cli automates scaffolding, eval, and deployment on top. If you prefer direct control or non-Google Cloud deployment, use ADK standalone with gcloud CLI.

LangChain + LangSmith (with custom Google Cloud CI/CD)

LangChain is framework-agnostic; LangSmith provides eval and monitoring. Requires manual integration with Google Cloud deployment pipelines. Better for non-Google cloud or multi-cloud scenarios.

CrewAI + manual evaluation and deployment

CrewAI offers agent orchestration with minimal boilerplate. No integrated eval or deployment; you manage those separately. Suitable if you want lightweight agent framework without Google Cloud lock-in.

Software development agency

Build on agents-cli with DEV.co software developers

Start with `uvx google-agents-cli setup` or integrate skills into your coding assistant. Review the quickstart tutorial and explore eval and deployment workflows in the full documentation.

Talk to DEV.co

Related open-source tools

Surfaced by semantic similarity across the DEV.co open-source index.

cline adk-python better-agents

Related on DEV.co

Explore the category and the services that help you build with it.

AI Frameworks AI Application Development Cloud Deployment & DevOps API Development Talk to DEV.co

agents-cli FAQ

Do I need a coding assistant to use agents-cli?

No. The CLI works standalone from your terminal (`agents-cli scaffold`, `agents-cli eval`, `agents-cli deploy`). Skills are optional and benefit coding assistants; the core CLI is independent.

Can I use agents-cli with an existing ADK project?

Yes. `agents-cli scaffold enhance` adds deployment, CI/CD, and RAG infrastructure to existing projects. Full scaffolding is not required; you can adopt agents-cli incrementally.

Do I need Google Cloud for local development?

No. You can scaffold, run, and evaluate agents locally using an AI Studio API key (free Gemini access). Google Cloud is required only for cloud deployment and features like Agent Runtime.

What is the difference between agents-cli and ADK?

ADK is the agent framework (APIs, runtime, tools). agents-cli is a CLI and skill suite that scaffolds projects, generates ADK code patterns, evaluates agents, and deploys to Google Cloud—automating the end-to-end workflow.

Software developers & web developers for hire

Need help beyond evaluating agents-cli? DEV.co is a software development agency offering software development services and web development for teams of every size. Our software developers and web developers build custom software, web applications, APIs, and ai frameworks integrations — and maintain them long-term.

Ready to Build Enterprise AI Agents?

Start with `uvx google-agents-cli setup` or integrate skills into your coding assistant. Review the quickstart tutorial and explore eval and deployment workflows in the full documentation.

Talk to DEV.co Schedule a Consultation