ADK is a code-first TypeScript toolkit for building, evaluating, and deploying sophisticated AI agents with flexibility and control.
The toolkit addresses the need to define agent behavior, orchestration, and tool use directly in code rather than through configuration or visual builders. It provides full type safety through TypeScript and Zod schema validation, enabling robust debugging and versioning. Agents can be composed into sequential, parallel, loop, and routed workflows, and can delegate to remote agents via the A2A protocol. The framework ships with built-in tools for Google Search, Google Maps, Vertex AI Search, and URL context, while also supporting MCP servers and custom function wrapping.
The project suits teams building multi-agent systems who want to work in TypeScript with compile-time type safety and the ability to run agents in both Node.js and browser environments. It includes a development UI for interactive testing and debugging, plus CLI commands for scaffolding, local testing, and deployment to Cloud Run. The toolkit supports both Google AI Studio API keys and Vertex AI authentication, making it flexible for different deployment contexts.
The project receives issues from both core maintainers and external users, indicating adoption beyond the immediate team without an overwhelming support burden. Maintainers typically respond to new issues and pull requests within hours. Work in the issue tracker centers on bug fixes, enhancements, and code review activities.
