Grade Your AI Prompt in Seconds
Paste any prompt. Get a letter grade plus specific fixes across 7 dimensions that predict output quality. No login. No API. Nothing sent anywhere.
Try a sample
Overall grade
—/100
Paste a prompt to begin
How the 7 dimensions work
Each dimension is a proven quality signal drawn from the 19 Laws of Prompting, the 32 ChatGPT Cheat Codes, and testing across ChatGPT, Claude, Gemini, and Perplexity. Pattern matching detects each signal; weights reflect real-world impact on output quality.
The scorer is a directional guide, not an oracle. A prompt can score 95 and still miss for your specific workflow, and a 75 can ship fine if your workflow has tight constraints elsewhere. Use it to catch gaps, not as a stamp of approval.
Scoring weights
- Task clarity (22%) — the single biggest driver of output quality.
- Context (20%) — who, what, why; prevents generic output.
- Output format (15%) — the difference between usable and rework.
- Constraints (15%) — tone, audience, word count, avoid-list.
- Role (10%) — shifts style and vocabulary meaningfully.
- Examples (10%) — few-shot pattern locking.
- Verification (8%) — the hallucination guard.
Keep leveling up
Framework
19 Laws of Prompting
The foundational rules behind every dimension in this scorecard.
Read the lawsCopy-ready
32 ChatGPT Cheat Codes
Prompt modifiers you can paste into any chat to instantly improve output.
See the codesDebugging
7 Costly Prompt Mistakes
The exact failure modes this scorecard catches — with before/after examples.
Fix the mistakesFrequently asked questions
How does the Prompt Scorecard work?+
It runs entirely in your browser — no server, no API call, no data sent anywhere. Your prompt is analyzed against 7 dimensions proven to predict output quality: task clarity, context, output format, constraints, role, examples, and verification. Each dimension gets a 0–100 score, and an overall weighted grade (A–F) is calculated.
Is my prompt sent to any AI model or stored anywhere?+
No. The scoring is pure JavaScript pattern matching running client-side. Nothing is logged, stored, or sent over the network. You can disconnect from the internet after loading this page and it will still work.
Why these 7 dimensions?+
They come from the 19 Laws of Prompting and real-world testing across ChatGPT, Claude, Gemini, and Perplexity. Task clarity and context together account for most output quality. Constraints and output format are where professional prompts separate from beginner prompts. Examples and verification are the polish layer.
My prompt scored low — now what?+
Each dimension with a sub-80 score shows a specific fix suggestion with a link to the exact framework or cheat code that addresses it. Apply the fixes, rescore, and watch the grade climb.
Does the scorecard work for Claude, Gemini, and other models?+
Yes. The 7 dimensions apply across every major model — GPT-5, Claude 4.6, Gemini 3.1, Perplexity, and open-source models. Model-specific tweaks matter less than prompt structure.
Can I use this for coding prompts or image generation prompts?+
Yes, with caveats. The scorer is tuned for text generation workflows. Coding prompts benefit most from task clarity, context, and examples. Image prompts score slightly differently — for those, constraints and format carry more weight, role less.