Solo — design & full-stack
Live · 2026Regression Reader
A regression testing tool that improves efficiency and removes distractions.
- TypeScript
- Claude Code
- UI/UX
- Testing
During my summer 2026 internship at Tektonux, a user experienced focused software development company, one of my tasks was to run regression tests. During my first test I noticed how cumbersome and mentally taxing running these tests were. The issues with the test arose from two main problems.
The first was, after a few minutes of testing, the plain text test instruction document would begin to swim before my eyes. I would lose which line I was on and sometimes accidentally skip steps. The second problem was I had a separate document where I recorded notes and bugs. Most bugs were tied to a specific line in the test plan and reconnecting the two after the fact was tedious. To solve these problems I built a regression testing tool with Claude code.
Compare the two screenshots below, step 6 is easy to lose in the original, impossible to lose in my tool.
To fix the first problem, where I struggled to keep track of which step I was on. The Regression Reader locks your attention to a single line at a time by highlighting it, the current line stays highlighted till you arrow down to the next one. Nothing else is competing for your attention.
Even five minutes into a regression test, that highlight bar is a life saver.
The second problem was juggling two documents at once. So in the Regression Reader, you write your notes inside the test plan itself instead of a separate file. Press a key on any line and a color-coded note bubble opens right there — orange for a general note, red for a bug, purple and green for whatever else you want to track.
The note is permanently tied to that exact line. A colored dot signifies a line with a note. This makes going over the bugs you discovered effortless.
Once bugs have been logged and annotations have been added, Summary Mode collapses the whole plan down to just the lines with notes, filterable by note color. It turns hours of scrolling into a short list of exactly what needs attention. Use the space bar to expand context.
Writing up results after a regression run takes a fraction of the time it used to.
I built the first working version in one evening — about two hours with Claude Code — then brought it to work the next day and ran my next regression pass on it. It held up.
A teammate started using it not long after, and the plan going forward is for more of the team to pick it up and keep extending it. I went looking for a fix to my own problem and ended up with something my team at Tektonux is actually using.
More projects