메뉴
Standards and constraints for the unified google-genai library.
Codex 또는 Claude로 설치 이 Prompt를 복사해 Codex, Claude 또는 다른 어시스턴트에 붙여 넣으면 Skill 페이지를 검토하고 설치를 진행할 수 있습니다.
SOC 직업 분류 기준
Expert in building, evaluating, and deploying AI agents using the Google ADK Python framework. Use this skill for writing agent logic, configuring multi-agent systems, and implementing tool integrations.
Advanced manipulation of text streams using the classic Unix power tools: awk and sed.
The discipline of writing robust, re-runnable system administration scripts that converge to a desired state.
Expertise in creating delightful, low-friction command-line interfaces using modern TUI principles.
Ensuring that code is well formatted
The art of identifying the root cause of software defects through heuristic analysis and logical deduction.
| name | google-genai-sdk-unified-v157 |
| description | Standards and constraints for the unified google-genai library. |
Standards and constraints for the unified google-genai library.
google-generativeai library is forbidden.client.aio.models.generate_content(...) for async operations.google-genai SDK, assume version 1.57 syntax.The user is utilizing the **Unified SDK**. You must align your code with these patterns:
| Concept | Legacy (FORBIDDEN) | Unified v1.57 (REQUIRED) |
| :--- | :--- | :--- |
| **Package** | `import google.generativeai` | `from google import genai` |
| **Client** | `genai.configure(...)` | `client = genai.Client(...)` |
| **Async** | `genai.ChatSession` | `client.aio.models.generate_content(...)` |
| **Methods** | `model.supported_methods` | `model.supported_actions` |
| **Config** | `generation_config={...}` | `config=types.GenerateContentConfig(...)` |
You are strictly forbidden from using `google.generativeai`. You must use the
unified `google-genai` library. Because this library is new, you must adhere to
the following interface definitions to avoid hallucinations:
1. **Import:** `from google import genai`
2. **Client:** `client = genai.Client(api_key=...)`
3. **List Models:**
```python
# CORRECT
for model in client.models.list():
print(model.name)
print(model.supported_actions) # NOT supported_methods
```
4. **Generation:**
```python
response = client.models.generate_content(
model='gemini-3-flash-preview',
contents='Hello'
)
print(response.text)
```
5. **Types:** Use `from google.genai import types` for type hinting config
objects.