Grading Rubric Builder Prompt
This prompt builds complete grading rubrics for any assignment: student-friendly criteria, four performance levels with observable descriptors ('cites 3+ pieces of evidence,' not 'good evidence'), point values, a 'what an A looks like' summary, and a self-assessment column. Fill in the assignment and standard — get a clean table back.
Clear, student-friendly rubrics for any assignment — criteria, levels, and descriptors in one pass.
Act as an assessment specialist. Build a grading rubric with these specs:
ASSIGNMENT: [e.g. 5th grade persuasive essay, 3 paragraphs]
STANDARD: [e.g. CCSS.ELA W.5.1]
CRITERIA: [e.g. 4 — or let the AI propose]
LEVELS: [e.g. 4: Exceeding / Meeting / Approaching / Beginning]
POINTS: [e.g. 20 total]
Rules: criteria names in student-friendly language (no jargon); each cell describes OBSERVABLE performance ('cites 3+ pieces of evidence') not vague quality ('good evidence'); the top level is aspirational but achievable; include a one-line 'what an A looks like' summary at the top; add a student self-assessment checkbox column. Format as a clean table.
See what this prompt creates
A rubric is a contract: it tells students exactly what excellence looks like before they start, and it tells you exactly what you're grading after they finish. Bad rubrics — the vague "excellent / good / fair / poor" grids — do neither; they just launder subjectivity through a table. This prompt builds the good kind, because its core rule forces every cell to describe observable performance rather than quality adjectives. "Cites 3+ pieces of relevant evidence" is gradeable; "uses good evidence" is a mood.
The most underrated feature is the student self-assessment column. Having students score themselves against the rubric before submitting does two things: it makes them actually read the criteria (the eternal struggle), and the gap between their self-score and yours becomes the most productive five minutes of your feedback conference. The prompt includes it by default — don't delete it.
Getting the best result
- Cap criteria at 4–5. More criteria feel thorough and grade like punishment — for you and the student. If everything matters, nothing does; pick the load-bearing skills.
- Write the assignment description fully. "Persuasive essay" is thin; "5th grade persuasive essay, 3 paragraphs, on school uniforms" lets the descriptors reference the actual task.
- Check the top level for achievability. "Exceeding" should describe the best work a strong student could actually produce — not a graduate thesis. If no student could hit it, it's demotivating decoration.
- Share it before they start. A rubric delivered with the assignment prevents problems; a rubric revealed at grading just documents them.
3 variations to try
Single-point rubric: add "single-point format: one column of success criteria with blank space for strengths and growth areas" — less grading, richer feedback, increasingly popular.
Peer-review version: "rewrite the descriptors as peer-feedback prompts ('Does the essay cite 3+ pieces of evidence?')" for structured peer assessment days.
Standards-based: "levels labeled 4/3/2/1 with proficiency language, no points" for standards-based grading systems.
How to use this prompt
- Copy it — hit the copy button above to grab the raw prompt text.
- Fill the brackets — replace anything in [BRACKETS] with your own details.
- Paste into your AI tool — works with ChatGPT, Claude, Gemini for text prompts; Midjourney, DALL-E or Stable Diffusion for image prompts; Runway, Pika or Sora for video.
- Iterate — tweak one bracket at a time until the output is exactly right.
Frequently asked questions
How many criteria should a rubric have?
Four to five. More feels comprehensive but grades like punishment — for you and the student. Pick the load-bearing skills the assignment is actually about; everything else is comments-section material.
What's wrong with 'excellent / good / fair / poor' descriptors?
They describe quality, not performance — 'good evidence' means something different to every reader, including you on a tired Sunday. Observable descriptors ('cites 3+ pieces of relevant evidence') are gradeable, defensible, and teachable.
Should students see the rubric before the assignment?
Always. A rubric shared upfront is a roadmap; revealed at grading time it's just an autopsy report. The self-assessment column in this prompt only works if students have the rubric while they work.
Can I convert this to standards-based grading?
Yes — use the standards-based variation: 4/3/2/1 proficiency levels, no points. The observable-descriptor structure ports directly; only the level labels and math change.
