GoogleCloudPlatform/agent-starter-pack

Features

• Agent-to-Agent Communication - Links agents built with different frameworks using a standardized communication protocol for interoperability.
• Template-Based Launches - Deploys ready-made agent templates for common patterns like reasoning-and-acting and retrieval-augmented generation.
• Agentic Reasoning Frameworks - Creates a conversational agent that reasons through problems, selects tools, and takes actions.
• Customization Tools - Ships a CLI tool that modifies starter agent prompts, tools, data sources, and reasoning logic.
• Starter Agent Libraries - Provides a library of pre-built, framework-specific agent templates for selection.
• Multi-Agent Orchestrators - Coordinates multiple specialist agents through agent-to-agent protocols to complete complex tasks.
• Multi-Agent System Orchestration - Enables distributed agents built with different frameworks to communicate using a standardized protocol.
• RAG Data Pipelines - Connects data ingestion and embedding pipelines to search or vector services for retrieval-augmented generation.
• Retrieval-Augmented Agents - Ships a template for creating agents that answer questions by retrieving relevant information from indexed documents.
• Conversational Agents - Provides a minimal ReAct agent template for bootstrapping general-purpose conversational agents.
• ReAct Deployments - Deploys an agent that iteratively calls tools to reason through and answer user queries.
• Document Q&A - Ingests custom documents, generates embeddings, and retrieves context to answer user questions.
• Document Question Answering - Implements retrieval-augmented generation with a data ingestion pipeline for document question answering.
• Agent Execution Traces - Traces every agent operation with distributed tracing spans and exports them to a cloud trace service.
• Cloud Provisioning Tools - Ships a tool that provisions cloud infrastructure for agent deployment using declarative infrastructure-as-code.
• Document Ingestion Pipelines - Ingests documents into search or vector datastores via cloud storage uploads or pipeline runs.
• Search and Vector Service Selectors - Provides configuration for selecting and provisioning search or vector datastores for document ingestion.
• Multi-Format Document Ingestion - Ingests and normalizes unstructured files, structured data, and custom JSON for downstream processing.
• Search and Vector Service Selectors - Provides configuration for selecting and provisioning search or vector datastores for document ingestion.
• GenAI Project Scaffolders - Creates new generative AI application projects from templates with infrastructure, CI/CD, and observability.
• Agent Scaffolding - Provides a minimal, production-ready template for a reasoning-and-tool-use agent ready to deploy.
• GenAI Agent Scaffolders - Creates production-ready agent projects from templates in under a minute using a single command-line invocation.
• Production-Ready Agent Scaffolders - Generates complete project structures with backend, optional frontend, and deployment infrastructure from templates.
• Bootstrapping Tools - Provides a single-command tool that bootstraps full CI/CD pipelines with infrastructure and repository integration.
• Agent CI/CD Bootstrappers - Sets up complete CI/CD pipelines for agent projects with a single command.
• Agent Project Enhancers - Adds deployment infrastructure, CI/CD pipelines, and observability to existing agent projects.
• Bootstrapping Tools - Provides a single-command tool that bootstraps full CI/CD pipelines with infrastructure and repository integration.
• Cloud Build and Platform Configurators - Configures complete build-and-deploy pipelines for cloud build services or CI/CD platforms with one command.
• Full Pipeline Bootstrappers - Creates complete CI/CD pipelines including cloud project and remote repository configuration in one operation.
• Runner-Aware Pipeline Setup - Automatically detects CI/CD runners and configures infrastructure with remote repository integration.
• Staged Agent Deployments - Automates full deployment lifecycle with staging and production environments requiring manual approval.
• Cloud Agent Deployers - Provisions infrastructure and deploys agents to managed runtimes with a single command.
• Production Deployments - Deploys GenAI agents to production with infrastructure-as-code, CI/CD pipelines, and automated rollback capabilities.
• Cloud Provisioning - Applies infrastructure-as-code scripts to create cloud resources with remote state management.
• Live Reload Playgrounds - Launches a local development environment with live reload for testing and iterating on agent behavior.
• Remote Template Generators - Generates deployable agent projects by fetching and merging templates from remote Git repositories.
• GenAI Application Scaffolders - Creates new generative AI application projects from templates with infrastructure and configuration.
• Local Agent Simulation - Launches an interactive local playground with live reload for testing agent behavior during development.
• Codebase Injectors - Injects agent features into an existing codebase without creating a new project directory.
• Real-Time Streaming - Handles real-time audio and video input and output for interactive agent experiences.
• Personalized Tutoring Agents - Provides a template for building an educational agent that adapts content to individual learning styles.
• Personalized Tutoring Agents - Provides a template for building an educational agent that adapts content to individual learning styles.
• Multimodal Streams - Handles audio, video, and text inputs and outputs with low-latency bidirectional streaming.
• Cloud Storage Sync Triggers - Triggers automatic or manual synchronization of documents from cloud storage into datastores.
• Ingestion Schema Definitions - Defines source file formats for correct parsing and indexing during document ingestion.
• Agent Capability Enhancers - Adds agent capabilities to an existing project without creating a new directory structure.
• Graph-Based Templates - Provides a template with explicit state management and complex reasoning flows for structured agent behavior.
• Multilingual Starter Templates - Provides starter templates in several programming languages for agent development.
• Development Environment Provisioners - Provisions standalone development environments with infrastructure-as-code for agent testing without full CI/CD.
• Agent Template Packagers - Packages custom agent code, dependencies, and infrastructure into reusable Git repositories.
• Boilerplate Upgraders - Updates projects to the latest template version without overwriting user-made modifications.
• Agent Template Catalogs - Displays a catalog of pre-built agent types and templates for starting new projects.
• Project Version Updaters - Updates existing agent projects to the newest version of the templates and tooling.
• Runner Detection and Configuration - Detects and configures CI/CD runners based on project structure.
• Existing Project CI/CD Injectors - Injects CI/CD workflows and infrastructure-as-code into pre-existing codebases.
• Existing Project CI/CD Injectors - Injects CI/CD workflows and infrastructure-as-code into pre-existing codebases.
• Standalone Dev Deployments - Provisions standalone infrastructure and deploys agents without a full CI/CD pipeline.
• Version Upgrades - Updates existing projects to the latest version of the agent template and its dependencies.
• Chat Platform Connectors - Provides a connector to link Vertex AI agents to Google Chat for direct user interaction.
• Multi-Agent Fact-Checkers - Critiques user statements using a multi-agent system with language models and web search grounding.
• CI/CD-Integrated Workflows - Coordinates specialist agents with CI/CD gates and infrastructure-as-code for enterprise deployments.
• Data Warehouse Logging - Logs detailed agent interactions including model calls and tool outcomes to data warehouses for analysis.
• Ticket and Rollback Triggers - Detects issues in logs and rollouts, then automatically creates tickets or triggers rollbacks.
• AI-Powered Monitors - Uses an AI agent to watch system health, detect anomalies, and provide intelligent insights for production environments.
• Automated Remediation Analyzers - Investigates log entries, automatically creates prioritized tickets, and triggers rollbacks for detected issues.
• AI-Enhanced Monitors - Helps DevOps teams track system health, detect anomalies, and generate intelligent insights for production environments.
• Agentic Frameworks and LLM Tools - Production-ready templates for building generative AI agents.

Istoric stele