Loading page
About this tool
Build an AI prompt that generates a weighted grading rubric with named criteria, performance levels and exact point bands.
The Rubric Prompt Builder generates an AI prompt that produces an analytic grading rubric: weighted criteria as rows, performance levels as columns, and observable descriptors in every cell. Criterion weights are converted into whole point maxima with largest-remainder apportionment, so a 100-point rubric weighted 30/30/25/15 always allocates exactly 30, 30, 25 and 15 points. Teachers and course designers get a rubric prompt with fixed numbers the AI cannot drift from.
Open Rubric Prompt Builder on AltFTool — it loads instantly in your browser.
Add your input to the workspace.
Adjust the options until the result looks right.
Copy or download the output and put it to work.
Weights become whole point maxima that sum exactly to your total — no rounding surprises at grading time.
The prompt bans vague adjectives and demands measurable evidence in every cell, at every level.
Choose 3, 4 or 5 performance levels with conventional education labels rather than inventing your own.
An analytic rubric scores each criterion separately on its own row, unlike a holistic rubric that gives one overall score. Analytic rubrics are the standard for transparent grading because students see exactly where points were earned or lost, and this tool fixes each row's maximum points before the AI writes a single descriptor.
By largest-remainder apportionment: each weight percentage is turned into a fractional point quota, whole points are assigned by the floor first, and any leftover points go to the criteria with the largest fractional remainders. The per-criterion maxima therefore always sum exactly to the rubric total.
Three to five levels is standard practice; this tool offers exactly those options. Four levels (Exemplary, Proficient, Developing, Beginning) is the most common choice — it mirrors widely used academic rubric frameworks and avoids a safe middle option that graders overuse on odd-numbered scales.
Yes. The tool validates that weights sum to exactly 100% before generating a prompt and shows the current sum when they do not, because a rubric whose weights do not total 100% cannot be mapped cleanly onto a fixed point total.