This repository contains reusable skills that add relevant context to AI agents when they need to interact with the Gemini API. Each skill encapsulates best practices, SDK guidance, and domain-specific knowledge that helps agents produce more accurate API code. The skills are lightweight context modules that address the knowledge gap between when an LLM was trained and the current state of SDKs and best practices. Evaluations show significant improvements in agent accuracy when using these skills compared to base models.