Original prompts
Our growing collection of templates are original editorial drafts prepared for OnlyPrompts. They have not been run across the listed AI platforms. Platform labels indicate intended use, not verified compatibility.
How template scores work
All prompts receive an AI-assisted editorial assessment of their written instructions. Five dimensions—task clarity, input guidance, output structure, verification, and safeguards—each receive 0 (absent), 1 (partial), or 2 (explicit). The total is out of 10. A high score means the template is well specified, not that a model produced a better answer. Each review records its reasoning and date. The assessment uses a consistent checklist: a general source-check instruction earns partial verification credit; a specific comparison or audit earns explicit credit. Many templates have similar scores because they share safeguards. Scores do not rank popularity or predict results. We do not publish invented community votes or aggregate star ratings.
Linked libraries
Library descriptions are original summaries linked to the source. We do not redistribute repository contents. Stars are dated snapshots, not ratings or security guarantees. Check current licences and permissions at the source.
What “tested” will mean
A tested entry must record the exact input, model/version, date, actual output, evaluation criteria, and limitations. Until that evidence exists, it remains a draft. We do not invent outputs or model rankings.
Commercial independence
No affiliate or advertising programmes are currently active. Future commercial placements will be labelled. Recommendations and rankings should reflect reader usefulness rather than commission.
Corrections and submissions
Public submissions remain closed until moderation, contributor permissions, and a contact channel are configured. We review source links and claims as the library develops.