About PromptCaret

PromptCaret is built and run by Prateek Chaudhary, working alone, as a sole proprietor in India. There is no team page because there is no team. Questions, complaints and bug reports all reach the same person at support@promptcaret.com.

The seven things the scorer looks for — role, task, context, constraints, output format, examples and specificity — are not invented. They are the parts that show up consistently in the published system prompts of production AI tools: Cursor, v0, Devin, Claude and others, collected in the system-prompts-and-models-of-ai-tools corpus.

No model was fine-tuned on that corpus. It is used as a reference: the rubric was derived from it by hand, and the AI rewrite retrieves the most relevant passages from it at request time so a suggestion can point at the pattern it came from. That is a deliberate choice — it is cheaper, and it means every suggestion can be traced to something a real tool actually does.

The score counts whether the parts of a prompt are present. It is pattern matching, so it can see that you wrote a constraint; it cannot read whether the constraint was a sensible one. A hollow template that names every section and fills none of them scores 100, and that is the correct answer to the question the engine asks.

This is stated everywhere on the site rather than buried here, because a score that oversells itself is worse than no score. If you want to see the reasoning applied properly, the Academy works through one part per lesson.

Scoring is free, unlimited, and needs no account, and it stays that way — it costs nothing to run because it runs in your browser. The AI rewrite calls a paid API, so it is capped at three a day on the free tier and unlimited on a pass. Passes are one-time purchases. Nothing auto-renews, nothing is billed silently, and there is no subscription to remember to cancel. See pricing.