google/adk-java
An open-source, code-first Java toolkit for building, evaluating, and deploying sophisticated AI agents with flexibility and control.
What it solves
It provides a code-first toolkit for developers who need precise control and flexibility when building, testing, and deploying AI agents, particularly those that need to integrate deeply with Google Cloud services.
How it works
Developers define agent behavior, orchestration, and tool usage directly in Java code. The toolkit allows for the creation of single agents or modular multi-agent systems where specialized agents are organized into hierarchies. It supports integrating pre-built tools, custom functions, and OpenAPI specifications to extend agent capabilities.
Who it’s for
Java developers building sophisticated AI agents who prefer a programmatic approach over configuration-based setups for better debugging, versioning, and testability.
Highlights
- Code-First Approach: Logic and orchestration are defined in Java, enabling standard software engineering practices like versioning and testing.
- Modular Multi-Agent Systems: Ability to compose multiple specialized agents into scalable hierarchies.
- Rich Tooling: Integration with Google Search and other custom or OpenAPI-based tools.
- Development UI: Includes a built-in interface for testing, debugging, and showcasing agents.
- A2A Integration: Supports remote agent-to-agent communication via the A2A protocol.