prompt.kred
Now in early build

Prompt engineering,
minus the guesswork.

Every prompt scored for quality, priced across GPT-4o and Claude, and proven by people who've actually run it — a real library to search, not just a calculator.

Browse the library Try the evaluator

Cost, upfront

Token counts and real cost estimates across GPT-4o, Claude Sonnet, and Claude Haiku — know the bill before you run it, not after.

Quality scored

Every prompt gets an automatic quality check on structure, clarity, and specificity — no guessing if it'll actually work.

A real library

Reusable, parameterized prompts you can fill in and copy in seconds — searchable, taggable, and sorted by what actually works.

Favorite & fork

Save what you'll reuse, and fork anyone's prompt into your own library to tweak and make it yours.

Rated by real use

Star ratings and comments from people who actually ran the prompt — not just view counts.

Standalone evaluator

Paste any prompt — even one you'd never publish — and get the same cost and quality breakdown instantly.

How it works

1

Find or write a prompt

Search the library, or start from scratch in the evaluator.

2

See the real numbers

Token count, cost across models, and a quality score — instantly.

3

Copy and ship

Fill in any parameters, copy, and paste straight into your model of choice.

Ready to stop guessing?

Free to browse, free to publish, no API key required to get started.

Get started