The Agent Development Kit (ADK) has been extended from a Python‑centric offering to native SDKs for Go, TypeScript, Java, and Kotlin. All five language kits follow the same architectural blueprint and expose the same open protocols, so teams can select the runtime that matches their performance, ecosystem, or deployment preferences without introducing an extra Python service.
Unified Open Protocols
Each language implementation includes a client for the Model Context Protocol (MCP), enabling any ADK‑based agent to call a custom MCP server regardless of the language it runs in. For multi‑agent scenarios, the kit also implements the Agent2Agent (A2A) protocol, which standardizes HTTP‑based communication between agents written in different languages. The shared protocol layer means that interoperability is a matter of configuration rather than custom glue code.
Language‑Specific Runtime Considerations
Because the SDKs are built from the ground up for their respective runtimes, developers must account for language‑specific characteristics. A Go‑based agent can be compiled as a high‑concurrency binary suitable for Cloud Run, while a TypeScript agent runs in a browser tab and can invoke Chrome’s on‑device model. Java and Kotlin agents align with JVM environments, and Python agents retain the familiar interpreter model. Choosing a language therefore influences concurrency model, startup latency, and the surrounding ecosystem tooling.
Deployment Options and Operational Impact
The ADK CLI provides a one‑time setup command (uvx google-agents-cli setup) to verify Cloud authentication and install required tools. A scaffold command then generates a project skeleton with A2A endpoints and deployment descriptors for Cloud Run, Gemini Enterprise Agent Runtime, or Google Kubernetes Engine (GKE). For example, a Go agent can be scaffolded with:
agents-cli scaffold create my-go-agent \ --agent adk_go \ --deployment-target cloud_run \ --region us-central1These generated configurations reduce manual boilerplate, but operators still need to manage the chosen target’s lifecycle, scaling policies, and observability pipelines.
Related CloudNinjas coverage: Google Cloud.
What This Means For Practitioners
Practitioners should evaluate which language aligns with their performance and ecosystem needs, then use the ADK scaffold to generate consistent deployment artifacts. Pay attention to the runtime’s security posture—especially when exposing A2A HTTP endpoints—and integrate standard observability practices for the selected platform. Monitoring the evolution of the MCP and A2A specifications will be important as cross‑language agent ecosystems mature.
