Axiom

Overview

I teach an online quantitative reasoning course where students submit Excel-based discussion assignments every week. Writing thorough, consistent feedback for each submission was the most time-consuming part of grading, so I built Axiom: a Chrome extension that runs inside the university’s learning platform. While replying to a student, I type / and a keyword (or browse by assignment), and Axiom inserts a fully formatted response, filling in details such as the student’s name.

What I built

• The extension itself (Manifest V3): an in-page picker with slash-command search, browse-by-assignment mode, placeholder fill-in, and insertion that works with the platform’s rich-text editor (TipTap/ProseMirror).
• An audited feedback library. I analyzed all 14 assignment templates, including their hidden answer-key and self-checking logic, and grew the library from 114 to 165 responses, correcting dozens that pointed students to the wrong cells and documenting bugs in the templates themselves.
• Reliability and safety fixes: a data-loss bug that could wipe the library, doubled text insertion, keyboard capture on unrelated sites, and a privacy risk from pre-filling the previous student’s name.
• A safe update system. New and improved responses merge into each user’s library without overwriting their own edits or bringing back deleted items.
• Better submissions. I diagnosed why some submissions broke (unsupported spreadsheet programs, browser previews, altered templates) and turned the findings into clear feedback and course policy.

Impact

• A typical feedback response that used to take at least 5 minutes per student now takes about 5 seconds.
• More consistent, more specific feedback, with exact cells, formulas and expected answers.
• Used with nearly 1,000 students so far.

How it was built

Developed with AI assistance (Claude Code) for coding, testing and template analysis, with every change reviewed and approved by me.