Refactoring with Gemini in Android Studio (Part 2) – Planning Code Changes with Checkpoints

In this second episode about refactoring with Gemini in Android Studio, I show the prompt and AI skill I use to plan a refactoring before touching the code, and how I challenge the generated plan to get the results I actually want.

I'll show:
- Which prompt/skill structure I use to get a useful refactoring plan.
- How I guide the AI toward a plan that fits the actual codebase.
- When and how I push back on the generated plans before executing anything.
- What I consider essential for getting the best possible planning results from Gemini.

This episode continues the same project from Part 1, ComposablePreviewScanner, and focuses on the planning stage before refactoring. You can fin the project here:
🔗 https://github.com/sergio-sastre/ComposablePreviewScanner

🔗 Official Android Skills:
https://developer.android.com/tools/agents/android-skills?utm_campaign=deveco_gdemembers&utm_source=deveco

🔗 Original skill by Andrea Laforgia:
https://github.com/a4al6a/claude-code-agents/tree/main/alf-refactoring-advisor
⚠️ Beware it might differ from the one used in the video

#androiddev #geminiai #androidstudio

Видео Refactoring with Gemini in Android Studio (Part 2) – Planning Code Changes with Checkpoints канала Sergio Sastre