Prompt Score Calculator
A free, browser-based tool. Runs entirely in your browser — no sign up, nothing stored.
Your prompt
0 words · Rule-based, runs in your browser — nothing is uploaded. Don’t paste passwords, API keys or private data.
How to use this calculator
- Paste the prompt you'd send to an AI model into the box.
- Read the score out of 100 and the five dimension bars (clarity, context, output format, constraints, role & safety).
- Check the Strengths, Missing & fixes and Warnings panels.
- Copy the improved-prompt template, fill the bracketed placeholders, and use it in your AI tool.
How it works
The tool checks your prompt for the building blocks of a strong instruction: a defined role, a clear task, context, the intended audience, an output format, constraints, examples, success criteria and any request for sources. Each is detected with deterministic keyword and structure rules — no AI is involved.
Those signals feed five weighted dimensions — Clarity (30%), Context (20%), Output format (20%), Constraints (20%) and Role & safety (10%) — which combine into the overall score out of 100. It also flags weaknesses like a too-short prompt, vague wording (“something”, “good”), or contradictory length/tone.
The improved prompt is a structured template assembled from your text, not an AI rewrite — it lays out Task, Context, Audience, Output format, Constraints and Success criteria so you can fill any gaps before sending.
Worked example
“write something good” vs a structured prompt. “write something good” scores low — it has a task verb but no role, context, audience, format or constraints, and triggers short/vague warnings. Adding a role, the subject, the audience, a 5-bullet ~120-word format and a friendly tone pushes the same idea into the Excellent band.
Common mistakes
- Leaving out the output format — saying “in 5 bullet points” or “as a table, ~150 words” gives far more consistent results than no format at all.
- Vague wording like “something”, “good” or “stuff” — the model has to guess. Name the subject, audience and the must-haves.
- Contradictory instructions — asking for a “short but comprehensive” answer, or “formal but casual” tone, pulls the model in two directions.
Frequently asked questions
Does this use AI or send my prompt anywhere?
No. It runs entirely in your browser using rule-based checks — your prompt is never uploaded or sent to any AI model. Still, avoid pasting passwords, API keys or private data into any tool.
Is a higher score always better?
It means your prompt includes more of the elements that usually produce reliable answers. A simple question may not need every element — use the breakdown as a checklist, not a hard rule.
Does it write the prompt for me with AI?
No. The improved prompt is a template scaffold built from your own text with placeholders to fill. It's a teaching aid to structure your prompt before you send it to a real AI tool.
Which AI tools does this work for?
Any of them — ChatGPT, Claude, Gemini, Copilot and others. Good prompt structure is model-agnostic.
Related tools
Tip: Enter any known values to calculate the remaining results.
All calculations run in your browser. Your inputs are never saved or transmitted.