Gemini Guide

What it does

The Gemini Guide skill helps you craft detailed prompts and code examples for Google’s Gemini AI by providing access to relevant documentation and SDK patterns. It retrieves information from Google's official Gemini API documentation, GitHub codegen instructions, and optionally local cached documents. Importantly, this skill does not directly call the Gemini API; instead, it provides context and guidance to be used within another tool like Claude.

When to use it

• When you need help understanding how to use Google's Gemini API.
• To find up-to-date SDK patterns for building with Gemini in JavaScript or Python.
• If you are unsure of the correct code structure or function names when working with Gemini.
• To quickly reference common mistakes and corrections related to Gemini development.

Key capabilities

• Retrieves Google Gemini API documentation via WebFetch.
• Accesses GitHub codegen instructions for current SDK patterns.
• Can utilize locally cached Gemini documentation (if available at ~/Documents/google-gemini-context/).
• Provides a prioritized lookup workflow to ensure accurate and relevant information.
• Offers a list of common mistakes and corrections to avoid incorrect code or prompts.

Example prompts

• "How do I use the Gemini API in JavaScript?"
• "What's the latest way to generate text with Gemini using Python?"
• "Show me an example of how to call the Gemini API."
• “What are common mistakes when using the @google/generative-ai package?”

Tips & gotchas

• The skill relies on WebFetch for retrieving online documentation. Ensure your agent has access to a functional web browsing tool.
• Always check the "deprecated patterns" reference (references/deprecated-patterns.md) before suggesting code examples to avoid outdated or incorrect approaches.
• Local cached documents are optional and can speed up lookups if available at ~/Documents/google-gemini-context/.