> ## Documentation Index
> Fetch the complete documentation index at: https://docs.neosantara.xyz/llms.txt
> Use this file to discover all available pages before exploring further.

# Google ADK

> Bangun dan koordinasikan agen AI otonom menggunakan Google Agent Development Kit (ADK) dan Neosantara via LiteLLM.

[Google Agent Development Kit (ADK)](https://adk.dev/?utm_source=neosantara-docs\&utm_medium=referral) adalah framework open-source berbasis code-first untuk membangun, menguji, dan mengevaluasi agen AI otonom. ADK menyediakan wrapper model `LiteLlm` bawaan, memungkinkan agen ADK mengeksekusi model Neosantara menggunakan prefix `neosantara/<model>`.

## Pengaturan

Instal library Google ADK dan LiteLLM:

```bash theme={"theme":{"light":"ayu-dark","dark":"catppuccin-latte"}}
pip install -U google-adk litellm
```

<CodeGroup>
  ```bash Bash / zsh icon="terminal" theme={"theme":{"light":"ayu-dark","dark":"catppuccin-latte"}}
  export NEOSANTARA_API_KEY="nsk_your_api_key_here"
  ```

  ```env .env icon="file-code" theme={"theme":{"light":"ayu-dark","dark":"catppuccin-latte"}}
  NEOSANTARA_API_KEY=nsk_your_api_key_here
  ```

  ```powershell PowerShell icon="terminal" theme={"theme":{"light":"ayu-dark","dark":"catppuccin-latte"}}
  $env:NEOSANTARA_API_KEY="nsk_your_api_key_here"
  ```
</CodeGroup>

## Inisialisasi Agen Dasar

Gunakan kelas `google.adk.models.lite_llm.LiteLlm` dengan prefix provider `neosantara/` untuk menghubungkan model Neosantara ke agen ADK. Kredensial dibaca otomatis dari environment variable `NEOSANTARA_API_KEY`.

```python theme={"theme":{"light":"ayu-dark","dark":"catppuccin-latte"}}
import os
import asyncio
from google.adk.agents import Agent
from google.adk.models.lite_llm import LiteLlm
from google.adk.runners import InMemoryRunner

# Inisialisasi agen dengan model Neosantara
researcher = Agent(
    name="researcher_agent",
    model=LiteLlm(model="neosantara/gemini-3.8-flash"),
    instruction="Anda adalah asisten riset teknologi yang menyajikan data secara ringkas dan objektif."
)

runner = InMemoryRunner(agent=researcher)

async def main():
    events = await runner.run_debug(
        "Jelaskan keunggulan AI Gateway regional bagi developer di Indonesia.",
        quiet=True
    )
    for event in events:
        if event.is_final_response() and event.content:
            for part in event.content.parts:
                if hasattr(part, "text") and part.text:
                    print(part.text)

if __name__ == "__main__":
    asyncio.run(main())
```

## Eksekusi Tool Calling

Google ADK mendukung pemanggilan fungsi Python biasa sebagai tool agen. Tambahkan parameter `allowed_openai_params=["tools"]` pada instance `LiteLlm` untuk meneruskan skema fungsi:

```python theme={"theme":{"light":"ayu-dark","dark":"catppuccin-latte"}}
import os
import asyncio
from google.adk.agents import Agent
from google.adk.models.lite_llm import LiteLlm
from google.adk.runners import InMemoryRunner

def hitung_estimasi_token(teks: str) -> str:
    """Menghitung estimasi token dan biaya komputasi berdasarkan panjang teks."""
    estimasi_token = len(teks) // 4
    return f"Panjang teks: {len(teks)} karakter. Estimasi: {estimasi_token} token."

analyst = Agent(
    name="token_analyst",
    model=LiteLlm(
        model="neosantara/gemini-3.8-flash",
        allowed_openai_params=["tools"]
    ),
    instruction="Gunakan tool hitung_estimasi_token jika pengguna menanyakan kalkulasi ukuran teks.",
    tools=[hitung_estimasi_token]
)

runner = InMemoryRunner(agent=analyst)

async def main():
    events = await runner.run_debug(
        "Berapa estimasi token untuk teks sepanjang 1000 karakter?",
        quiet=True
    )
    for event in events:
        if event.is_final_response() and event.content:
            for part in event.content.parts:
                if hasattr(part, "text") and part.text:
                    print(part.text)

if __name__ == "__main__":
    asyncio.run(main())
```

## Alur Kerja Multi-Agent (Sub-Agents)

ADK mengorkestrasi hierarki multi-agent menggunakan parameter `sub_agents`. Agen utama mengarahkan pertanyaan ke sub-agent terspesialisasi:

```python theme={"theme":{"light":"ayu-dark","dark":"catppuccin-latte"}}
import os
import asyncio
from google.adk.agents import Agent
from google.adk.models.lite_llm import LiteLlm
from google.adk.runners import InMemoryRunner

# Sub-agent untuk penalaran analitis
analyst_agent = Agent(
    name="analyst",
    model=LiteLlm(model="neosantara/deepseek-v4.1-flash"),
    instruction="Lakukan penalaran komputasi dan analisis throughput sistem."
)

# Root agent pengatur alur
lead_orchestrator = Agent(
    name="coordinator",
    model=LiteLlm(model="neosantara/gemini-3.8-flash"),
    instruction="Koordinasikan pertanyaan teknis. Delegasikan analisis mendalam ke sub-agent analyst.",
    sub_agents=[analyst_agent]
)

runner = InMemoryRunner(agent=lead_orchestrator)

async def main():
    events = await runner.run_debug(
        "Hitung efisiensi konkurensi 1000 RPM pada tier Basic.",
        quiet=True
    )
    for event in events:
        if event.is_final_response() and event.content:
            for part in event.content.parts:
                if hasattr(part, "text") and part.text:
                    print(part.text)

if __name__ == "__main__":
    asyncio.run(main())
```

## Langkah Berikutnya

* [Panduan Model Context Protocol (MCP)](/id/agents/overview)
* [Pilihan Model dan Harga](/id/gateway/models)
* [Integrasi LiteLLM Native](/id/integrations/litellm)
* [Integrasi CrewAI Multi-Agent](/id/integrations/crewai)


## Related topics

- [LiteLLM](/id/integrations/litellm.md)
- [CrewAI](/id/integrations/crewai.md)
- [Agno](/id/integrations/agno.md)
- [Katalog Model](/id/gateway/models.md)