AI Frameworks · GoogleCloudPlatform

agent-starter-pack

Agent Starter Pack is a Python framework from Google Cloud that provides production-ready templates for building GenAI agents on Google Cloud. It bundles infrastructure, CI/CD, observability, and deployment automation so teams can focus on agent logic rather than boilerplate.

Talk to DEV.co Discuss Implementation

Source: GitHub — github.com/GoogleCloudPlatform/agent-starter-pack

6.5k

GitHub stars

1.5k

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	GoogleCloudPlatform/agent-starter-pack
Owner	GoogleCloudPlatform
Primary language	Python
License	Apache-2.0 — OSI-approved
Stars	6.5k
Forks	1.5k
Open issues	46
Latest release	v0.41.3 (2026-04-25)
Last updated	2026-06-30
Source	https://github.com/GoogleCloudPlatform/agent-starter-pack

What agent-starter-pack is

A Python package offering pre-built agent templates (ReAct, RAG, multi-agent, ADK, LangGraph) with Terraform infrastructure-as-code, Cloud Run/Agent Engine deployment targets, Vertex AI evaluation integration, and CI/CD pipelines (Cloud Build and GitHub Actions). Built on Google Cloud services (Vertex AI, Cloud Run, Firestore).

Quickstart

Get the agent-starter-pack source

Clone the repository and explore it locally.

terminalbash

1git clone https://github.com/GoogleCloudPlatform/agent-starter-pack.git2cd agent-starter-pack3# follow the project's README for install & configuration

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

Best use cases

Fast prototype-to-production GenAI agents on Google Cloud

Teams needing to ship agent-based systems quickly can use pre-built templates, evaluation tooling, and infrastructure code to reduce time from concept to production deployment on Cloud Run or Agent Engine.

RAG and document-retrieval agents

The `agentic_rag` template integrates with Vertex AI Search and Vector Search, enabling rapid development of Q&A and retrieval-augmented generation agents without custom infrastructure design.

Multi-model agent orchestration and A2A interop

The ADK and ADK A2A templates support Agent-to-Agent protocol for distributed agent communication, suited for microservices-based agent architectures and cross-system integration.

Implementation considerations

Requires Python 3.10+, Google Cloud SDK, Terraform, and Make; standard CI/CD environment setup needed before initial project creation.
Templates are opinionated (Cloud Run, Firestore, Vertex AI); customization requires understanding of GCP service architecture and Terraform.
Evaluation integration uses Vertex AI evaluation service; costs and model availability depend on GCP regional quotas and billing configuration.
ADK agent template is recommended; LangGraph alternative available for teams already invested in LangChain ecosystem.
CI/CD setup is semi-automated; Cloud Build or GitHub Actions integration still requires repository and GCP project credentials.

When to avoid it — and what to weigh

Locked into non-Google Cloud providers — Starter Pack is GCP-first; templates and infrastructure code target Cloud Run and Agent Engine. Using AWS, Azure, or on-prem as primary targets requires significant rework.
Requires minimal dependencies or lightweight deployments — The framework bundles Python, Terraform, Make, and Cloud SDK dependencies. Projects needing ultra-lean or specialized runtime stacks may find it prescriptive.
In active maintenance mode only; no new features planned — README states ASP is in maintenance mode; active development moved to `agents-cli`. New projects should evaluate agents-cli; ASP receives critical fixes only.
Need for proprietary or non-Apache-2.0 licensing flexibility — Licensed under Apache-2.0, which is permissive for commercial use but includes patent termination clauses. Review legal team's IP policy if concerned.

License & commercial use

Licensed under Apache License 2.0 (Apache-2.0), an OSI-approved permissive open-source license. Permits commercial use, modification, and distribution with attribution and no liability.

Apache-2.0 permits commercial use without restriction. No proprietary restrictions on code produced; however, review Apache-2.0's patent termination clause (Section 3) with legal counsel if IP indemnification is critical. No vendor lock-in clause, but GCP service costs apply at runtime.

DEV.co evaluation signals

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

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

Security considerations

Infrastructure templates use Terraform and Cloud Run best practices (service accounts, least privilege). Observability tooling integrates with Cloud Logging and Cloud Trace. No explicit security audit or threat model published. Agent templates should be reviewed for prompt injection and data handling. GCP service-level security posture inherited from Cloud Run and Vertex AI; assess per deployment requirements.

Alternatives to consider

agents-cli (Google)

Official successor to Agent Starter Pack. Unified CLI, bundled ADK coding skills, end-to-end lifecycle tooling, and first-class Google Cloud Agent Platform support. Recommended for new projects.

LangChain LangGraph (LangChain)

Multi-cloud agent orchestration framework with broader LLM provider support (OpenAI, Anthropic, etc.). Starter Pack includes LangGraph template, but LangGraph alone offers more flexibility outside GCP.

Amazon Bedrock Agents (AWS)

AWS-native agentic framework with Aurora datasources and Lambda integrations. Better fit for teams standardized on AWS; Starter Pack is GCP-first.

Software development agency

Build on agent-starter-pack with DEV.co software developers

Use `uvx agent-starter-pack create` to scaffold a fully functional agent with backend, frontend, and GCP infrastructure. Or evaluate agents-cli for new projects with enhanced tooling.

Talk to DEV.co

Related open-source tools

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

agents-cli generative-ai adk-python

Related on DEV.co

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

AI Frameworks AI application development Cloud deployment DevOps Talk to DEV.co

agent-starter-pack FAQ

Should I start a new agent project with Agent Starter Pack or agents-cli?

Start with agents-cli. ASP is in maintenance mode; agents-cli is the active evolution with new features, unified CLI, and first-class Agent Platform support. Migration from ASP takes minutes.

Can I use Agent Starter Pack outside Google Cloud?

Not easily. Templates and Terraform code target Cloud Run and Agent Engine. AWS, Azure, or on-prem would require significant customization; consider agents-cli or LangGraph for multi-cloud.

What LLMs are supported?

Templates use Gemini by default. ADK and LangGraph agents can be extended to support other models (OpenAI, Anthropic, etc.), but no out-of-the-box integration provided.

Is there a cost to using Agent Starter Pack?

The framework itself is free and open-source. GCP service costs (Vertex AI, Cloud Run, Vector Search, etc.) depend on agent usage and traffic; scaffold creates cost-optimized defaults.

Custom software development services

Adopting agent-starter-pack is usually one piece of a larger software development effort. As a software development agency, DEV.co provides software development services and web development expertise — pairing senior software developers and web developers with your team to design, build, and operate ai frameworks software in production.

Ship your first agent in 60 seconds

Use `uvx agent-starter-pack create` to scaffold a fully functional agent with backend, frontend, and GCP infrastructure. Or evaluate agents-cli for new projects with enhanced tooling.

Talk to DEV.co Schedule a Consultation