GPT Prompt Optimizer
A prompt underperforms when it omits what a model has to guess at. This finds those gaps, fills them with editable defaults, and tells you exactly what it added. No API call.
How the scoring works
It is a checklist, not an oracle. The score is a weighted sum across seven structural slots: a clear task (25), constraints (20), output format (15), context (15), audience (10), an example (10), and a role (5). Vague adjectives such as “good”, “professional” or “cutting-edge” are deducted, because they read as specific while specifying nothing.
Nothing is applied silently. Every slot the tool fills is listed, and each one can be toggled off. The defaults are deliberately generic, because the tool has no idea what your task actually is. Rewrite them.
No model is involved. This is a static rule set running in your browser. It cannot judge whether your idea is a good idea. It can only tell you that you left out the parts you had to leave out.