A library of engineered
prompts.

A curated archive · Est. 2026

Discover, score and run prompts across every major model. Every entry is benchmarked against gold-standard prompt engineering.

— The Method —

Search by meaning, not keywords. Vector embeddings surface prompts that resonate with your intent.

II.

Every entry is rated 0–100 against a rubric of clarity, specificity, structure and constraints.

III.

A multi-model playground compares latency, cost and output across GPT-5, Gemini and beyond.

Manifesto

"A prompt is a small instrument. Tune it well and it plays the model. Tune it poorly and the model plays you."

A library of engineeredprompts.