This repository provides production-ready Generative AI agent templates for Google Cloud, targeting developers building and deploying GenAI agents. It accelerates development by offering pre-built templates and addressing common challenges like deployment, evaluation, customization, and observability.

How It Works

The starter pack utilizes a command-line interface (CLI) to generate agent projects based on various templates, including ReAct, Retrieval Augmented Generation (RAG), multi-agent systems, and multimodal agents. It supports deployment to Google Cloud Run or the new Agent Engine, integrating with Vertex AI for evaluation and offering CI/CD automation and data pipelines for RAG.

Quick Start & Requirements

• Install: pip install agent-starter-pack
• Create Agent: agent-starter-pack create my-awesome-agent
• Prerequisites: Python 3.10+, Google Cloud SDK, Terraform.
• Documentation: Installation Guide, CLI Reference

Highlighted Details

• Offers pre-built agent templates: ReAct, RAG (Vertex AI Search/Vector Search), multi-agent (CrewAI), and live multimodal RAG (Gemini).
• Supports production-ready infrastructure with monitoring, observability, and CI/CD on Cloud Run or Agent Engine.
• Includes Vertex AI evaluation and an interactive playground for experimentation.
• Provides experimental CI/CD automation and Terraform-based data pipelines for RAG.

Maintenance & Community

• Contributions are welcome via the Contributing Guide.
• Feedback can be provided via GitHub issues or email at agent-starter-pack@google.com.

Licensing & Compatibility

• The repository is for demonstrative purposes and not an officially supported Google product.
• Usage of Google Cloud APIs is subject to Google Cloud Service Terms.

Limitations & Caveats

This repository is for demonstrative purposes only and is not an officially supported Google product. Users are responsible for resources deployed in their own Google Cloud projects.