Because feedback > grade

Sign in with Microsoft See how it works
Push to GitHub, or upload a document — graded the same way
Built by mafudge@syr.edu using Claude Code Grading IST356 since day one Python · pytest · GitHub or file upload Sign in with Microsoft
Who it's for

Two people, one moment

A student who isn't sure their work is right, and an instructor who can't tell forty of them individually in time for it to matter.

For students

Know exactly why. Then try again.

You think you're done. You're not sure it's right. Submitting used to mean waiting days to find out it wasn't.

  • Every criterion tells you what it took off for and what to do — "add a colon", "name the error types" — not just a number.
  • See which tests failed and what the grader saw.
  • Fix it, push, ask again. Keep going until it's right.
  • Submit however the assignment asks — push to GitHub, or upload a document.
  • It has read this assignment and this rubric — which a chatbot that has never seen either one hasn't.

In a demo course, one student went from 11/20 to 16/20 across four submissions, each one fixing what the feedback had flagged.

For instructors

Build it once. The feedback runs itself.

Write the assignment and the rubric once. Every submission after that is scored against it, in your words, without you writing the same comment forty times.

  • rubric.json holds your criteria, your levels, and your point values. It re-applies on every submission.
  • Automated checks score the objective cases — a sandboxed pytest run for code, exact counts for documents.
  • An AI judge scores the subjective ones against your levels, not its own.
  • Override any score, leave a note, finalize. The log shows who changed what, so a grade holds up when a student pushes back.

Grade less. Teach more.

For instructors

The trade

None of this is free. Here is what it costs you and what it returns.

What you invest
  • Time up front designing the assignment.
  • Writing the criteria, the point levels for each, and the hint a student sees when they miss it.
  • Keeping the rubric current when the assignment changes.
What you get back
  • Feedback in your words and to your standard, not a general model's.
  • The same standard applied to every student, the same way.
  • It arrives the moment they submit, while they can still act on it.
  • A record of every score, override and note, if a grade is questioned.
How it works

Push. Read. Fix. Push again.

From a student's first commit to a result they can act on — usually in minutes.

01

Submit your work

Push to GitHub, or upload a document — whichever the assignment asks for.

02

Checks and rubric run

Automated checks where they apply, then the AI judge against your rubric levels.

03

Feedback, per criterion

What was wrong, and what to do about it — on screen and by email.

04

Fix it and resubmit

Keep going until it's right. Your instructor can override any score.

Turn your next assignment into feedback

Sign in with your Microsoft account and connect a course. You write the rubric once; it does the grading from there.

Sign in with Microsoft See how it works