ADK Samples is a collection of sample agents built with the Agent Development Kit that demonstrates patterns for building AI agents.
The repository addresses the problem of starting agent development from scratch by providing working recipe examples that solve real problems. The collection is organized into two categories: canonical patterns in the core directory that teach focused concepts like OAuth flows, session memory, guardrails, and RAG patterns, and community-contributed recipes in the contrib directory that show broader use cases and industry-specific workflows. Each recipe includes its own setup and run instructions, allowing developers to fork examples as starting points for their own projects or study the patterns to understand how to structure agent implementations.
Developers should choose this collection if they are building customer service bots, research agents, or other AI-powered applications and want working foundations rather than starting from a blank page. The recipes support multiple language SDKs including Python, TypeScript, Go, Java, and Kotlin, so teams can select based on their existing technology stack. The project explicitly states that recipes are for demonstration and as starting points, not for production use, so they are best suited for learning and prototyping rather than direct deployment.
The project maintains a clear distinction between canonical patterns curated by the core team and community contributions, with documented contribution guidelines through a Recipe Checklist and Recipe Handbook. The repository includes tooling for recipe scaffolding and manifest generation within its agents skills directory to support consistent recipe development. The project accepts community contributions and provides GitHub issues as the support channel for questions and problems.
