Prompt Matrix Evaluator
Paste a prompt and get three things. A checklist of what it’s missing (task, output format, length, audience) that also flags vague words, contradictions, and leftover placeholders. One-click fixes for each gap. And two rewrites: a tightened copy with the filler removed, and a structured version with role, task, rules, and format laid out the way models follow most reliably.
How it works
- The checks are plain rules, not an AI judge. They catch the gaps that most often produce vague or rambling answers.
- Tightening only removes politeness and filler (“could you please”, “make sure to”, “thanks in advance”). Your actual request is never changed.
- The structured rewrite keeps your words and adds bracketed placeholders, like
[Who is this for?], wherever something is missing. Fill them in before you send it. - Add phrases your team overuses under Your own filler, and they’ll be removed too.
Token counts are estimates (about four characters per token).
Quick fixes
Each button edits your prompt and re-scores it.
Checklist
Structured rewrite
Tightened prompt
What was removed
What the savings are worth
Check your provider’s current pricing; this uses the number you enter.
Phrases to remove, one per line. Saved in this browser.