Prompt optimizer for Gemini
Rules traceable to official prompting guidance
Gemini wants Persona, Task, Context, Format, and Google explicitly warns against prompt tricks written for other models: Gemini 3 over-analyzes them. Its docs also recommend keeping temperature at 1.0 and requesting conversational output explicitly, because terse is the default.
What the rewrite does
- Restructures prompts into Google's Persona-Task-Context-Format frame
- Puts data first and anchors the trailing question, per Google's long-context guidance
- Strips chain-of-thought scaffolds that Gemini 3 over-analyzes
- Requests the verbosity you actually want, since terse-direct is the documented default
Nothing is invented: missing details become clearly marked placeholders, and we don’t store your prompts. 10,000 free words a month.