gemini

Gemini Command¶

The gemini command generates content using Google's Gemini AI models, providing powerful AI-assisted capabilities for code generation, explanations, and more.

Syntax¶

cursor-utils gemini [OPTIONS] PROMPT

Arguments¶

Argument	Description	Required	Example
`PROMPT`	The text prompt for Gemini to respond to	Yes	"Explain how promises work in JavaScript"

Options¶

Option	Description	Default	Example
`--model`	The Gemini model to use	`gemini-1.5-pro`	`--model gemini-2.0-flash`
`--temperature`	Sampling temperature (0.0-1.0)	`0.7`	`--temperature 0.9`
`--max-tokens`	Maximum tokens to generate	Based on model	`--max-tokens 1000`
`--system`	System instruction for guiding model behavior	None	`--system "You are a Python expert"`
`--format`	Output format (plain, markdown, json, rich)	`rich`	`--format markdown`
`--help`	Show command help	-	`--help`

Available Models¶

Model	Description	Best For
`gemini-1.5-pro`	Powerful general-purpose model	Default for most use cases
`gemini-2.0-pro-exp`	Experimental pro model	Complex reasoning, cutting-edge
`gemini-2.0-flash`	Faster, more efficient model	Quick responses, simpler tasks
`gemini-2.0-flash-exp`	Experimental flash model	Testing latest capabilities
`gemini-2.0-flash-thinking-exp`	Enhanced thinking capabilities	Step-by-step reasoning

Configuration¶

Before using the gemini command, you need to set up your Google Gemini API key:

cursor-utils config set gemini_api_key YOUR_GEMINI_API_KEY

You can obtain an API key from Google AI Studio.

Examples¶

Basic Usage¶

Generate a simple response:

cursor-utils gemini "Explain quantum computing in simple terms"

Code Generation¶

Generate a Python function:

cursor-utils gemini "Write a Python function to check if a string is a palindrome"

Using Different Models¶

Use a specific model:

cursor-utils gemini --model gemini-2.0-flash "Write a JavaScript function to sort an array"

Controlling Temperature¶

Lower temperature for more focused, deterministic responses:

cursor-utils gemini --temperature 0.2 "List 5 best practices for RESTful API design"

Higher temperature for more creative, varied responses:

cursor-utils gemini --temperature 0.9 "Generate unique names for a fantasy game"

Using System Instructions¶

Guide the model's behavior with system instructions:

cursor-utils gemini --system "You are a security expert specialized in code review" "Review this code for vulnerabilities: ..."

cursor-utils gemini --system "You are a technical writer who explains complex concepts clearly" "Explain OAuth 2.0 authorization flow"

Different Output Formats¶

Output in markdown format:

cursor-utils gemini --format markdown "Create documentation for a REST API endpoint"

Output in JSON format:

cursor-utils gemini --format json "List common HTTP status codes with descriptions"

Limiting Output Length¶

Specify maximum token generation:

cursor-utils gemini --max-tokens 500 "Summarize the history of artificial intelligence"

Use Cases¶

Code Development¶

Generate function implementations

cursor-utils gemini "Write a Python function to paginate API results"

Write test cases

cursor-utils gemini "Create unit tests for this function: def add(a, b): return a + b"

Debug code

cursor-utils gemini "Debug this code: for i in range(10): print(i[0])"

Documentation¶

Generate technical documentation

cursor-utils gemini --format markdown "Write documentation for a database connection class"

Create usage examples

cursor-utils gemini "Show examples of using the requests library in Python"

Learning¶

Explain concepts

cursor-utils gemini "Explain CORS and why it's important for web security"

Compare technologies

cursor-utils gemini "Compare MongoDB and PostgreSQL for web applications"

Advanced Techniques¶

Chaining Prompts¶

Building on previous responses for iterative development:

# First, generate a basic implementation
cursor-utils gemini "Write a Python function to convert Celsius to Fahrenheit" > temperature.py

# Then, improve the implementation
cursor-utils gemini "Improve this function to include error handling and type checking: $(cat temperature.py)" > temperature_improved.py

Structured Output¶

Request structured data for programmatic use:

cursor-utils gemini --format json "Convert this email to JSON with fields for sender, date, subject, and body: ..."

Redirecting Output¶

Save output to files:

cursor-utils gemini --format markdown "Write documentation for GraphQL mutations" > graphql-mutations.md

Best Practices¶

Be Specific: More specific prompts yield better results

# Less effective
cursor-utils gemini "Write a function"

# More effective
cursor-utils gemini "Write a Python function that validates email addresses using regex"

Consider Model Selection: Choose the appropriate model for your task
Use gemini-1.5-pro for complex reasoning, detailed responses
Use gemini-2.0-flash for quick, simple tasks
Tune Temperature: Adjust based on your need for creativity vs. determinism
Lower temperature (0.1-0.4) for factual, consistent responses
Higher temperature (0.7-0.9) for creative, varied responses

Use System Instructions: Guide the model's behavior using system instructions

cursor-utils gemini --system "You are a Python expert who writes well-documented, PEP 8 compliant code" "..."

Format for Readability: Use markdown for documentation, rich for interactive use

cursor-utils gemini --format markdown "Generate a tutorial on async/await in JavaScript"

Troubleshooting¶

API Key Issues¶

If you receive authentication errors:

Error: Invalid API key

Verify your API key is correctly set:

cursor-utils config get gemini_api_key

If it's missing or incorrect, set it:

cursor-utils config set gemini_api_key YOUR_GEMINI_API_KEY

Model Availability¶

If you receive an error about model availability:

Error: Model [model_name] is not available

Try using a different model:

cursor-utils gemini --model gemini-1.5-pro "Your prompt"

Rate Limiting¶

If you encounter rate limiting:

Error: Rate limit exceeded

Wait a few minutes and try again, or check your API usage limits.

Timeout Issues¶

For complex prompts that time out:

Error: Request timed out

Try simplifying your prompt or breaking it into smaller parts.