Smithery is now a part of Arcade.dev! Read more in our announcement here
Comprehensive toolkit for working with Google's Gemini API using the Google GenAI SDK...
Comprehensive toolkit for integrating Google's Gemini API using the google-genai Python SDK.
Use this skill to work with Gemini models for text generation, multimodal processing (images, audio, video, PDFs), structured outputs, and chat applications. The skill provides ready-to-use example scripts, API reference notes, and production-focused best practices.
uv add google-genai
from google import genai
client = genai.Client()
response = client.models.generate_content(
model="gemini-2.5-flash",
contents="Explain quantum computing simply",
)
print(response.text)
export GEMINI_API_KEY="your-key"Use for generating, summarizing, or transforming text.
Example scripts:
scripts/basic_generation.py - Simple text generationscripts/streaming_generation.py - Stream responses for better UXQuick example:
from google.genai import types
config = types.GenerateContentConfig(
temperature=0.7,
max_output_tokens=500,
)
response = client.models.generate_content(
model="gemini-2.5-flash",
contents=f"Summarize this text: {long_text}",
config=config,
)
Analyze images, extract text (OCR), or answer questions about visual content.
Example script: scripts/image_analysis.py
Quick example:
from PIL import Image
image = Image.open("photo.jpg")
response = client.models.generate_content(
model="gemini-2.5-flash",
contents=[image, "What's in this image?"],
)
Common tasks:
Extract structured JSON data from unstructured text using function calling.
Example script: scripts/function_calling.py
Quick example:
from google.genai import types
extract_person = {
"name": "extract_person",
"description": "Extract person info",
"parameters": {
"type": "object",
"properties": {
"name": {"type": "string"},
"age": {"type": "integer"},
"occupation": {"type": "string"},
},
"required": ["name"],
},
}
tools = types.Tool(function_declarations=[extract_person])
config = types.GenerateContentConfig(tools=[tools])
response = client.models.generate_content(
model="gemini-2.5-flash",
contents="John Doe is 30 and works as an engineer",
config=config,
)
function_call = response.candidates[0].content.parts[0].function_call
data = dict(function_call.args)
Build conversational applications with context management.
Example script: scripts/multimodal_chat.py
Quick example:
chat = client.chats.create(model="gemini-2.5-flash")
response = chat.send_message("Hello!")
print(response.text)
response = chat.send_message("What did I just say?")
print(response.text)
Stream long responses for better user experience.
Example script: scripts/streaming_generation.py
Quick example:
stream = client.models.generate_content_stream(
model="gemini-2.5-flash",
contents=prompt,
)
for chunk in stream:
print(chunk.text, end="", flush=True)
Choose the right model for your use case:
See the official model list for the latest stable and preview variants.
Process images, audio, video, and PDFs alongside text.
See: references/multimodal.md for comprehensive guide
Capabilities:
Enable structured outputs and tool use.
See: references/function_calling.md for detailed guide
Use cases:
Configure content safety settings per request.
Example:
from google.genai import types
config = types.GenerateContentConfig(
safety_settings=[
types.SafetySetting(
category=types.HarmCategory.HARM_CATEGORY_HATE_SPEECH,
threshold=types.HarmBlockThreshold.BLOCK_MEDIUM_AND_ABOVE,
)
]
)
response = client.models.generate_content(
model="gemini-2.5-flash",
contents=prompt,
config=config,
)
Common patterns:
import time
max_retries = 3
for attempt in range(max_retries):
try:
response = client.models.generate_content(
model="gemini-2.5-flash",
contents=prompt,
)
break
except Exception as exc:
status = getattr(exc, "status_code", None)
if status in (429, 503) and attempt < max_retries - 1:
time.sleep(2 ** attempt)
continue
raise
if response.prompt_feedback and response.prompt_feedback.block_reason:
print(f"Content blocked: {response.prompt_feedback.block_reason}")
Detailed guides available in references/ directory:
api_reference.md: Core SDK usage, models, parameters, authenticationbest_practices.md: Prompting, error handling, performance and cost tipsmultimodal.md: Image, audio, video, and PDF processing with examplesfunction_calling.md: Structured data extraction, tools, schemasReady-to-use scripts in scripts/ directory:
basic_generation.py: Simple text generationstreaming_generation.py: Streaming responsesimage_analysis.py: Image processing and analysisfunction_calling.py: Structured data extractionmultimodal_chat.py: Chat with multimodal inputsAll scripts include error handling and can be run directly or used as templates.
When to use which approach:
basic_generation.py patternstreaming_generation.py)function_calling.py)multimodal_chat.py)multimodal.md)Performance:
gemini-2.5-flash for best price-performancegemini-2.5-pro for quality and complex reasoninggemini-3-flash-preview or gemini-3-pro-preview when you want next-gen previewsmax_output_tokens to limit costsSecurity:
