About Commitscale
I built Commitscale because my agents were writing more code than I could read. Passing tests helped. I wanted to know what they were missing.
Hi, I'm John.
I spent six years at GitHub and Microsoft, leading support engineering teams and developer support organisations.
I've built funded and bootstrapped startups, including one I exited for an undisclosed amount. I care about helping developers improve the way they work.
Commitscale came from a problem in my own work. My coding agents were writing more code than I could review. The tests passed, but the same agent had written both the code and the tests. Both could miss the same mistake.
I built Commitscale to check that tests do their job. It tries small mistakes on a separate copy of the code, then runs the tests to see whether they notice.
We build Commitscale with coding agents, and use it to check our own changes. I still decide what needs my attention. Now I have more to go on than a passing test result.
The story behind the logo
We call it the Charter. It starts with a simple idea: you should be able to check what you're asked to trust.
Two halves that belong together.
The idea came from a medieval way of recording an agreement. The words were written twice on one sheet, then cut apart. Each person kept a half. To check the agreement, they put the halves back together. The cut and the letters had to match.
That kind of document is called a chirograph.
In our mark, the wave is the cut. The two short strokes echo the letters across it. The colours help you see two pieces of the same record.
We first chose it for Commitscale's record of each code change. For me, it also captures why we're building this: when you rely on a result, you should have something you can check.
How are you handling it?
If your agents are writing more code than you can read, I'd love to hear how you're handling it. What do you trust to the tests, and what do you still check yourself?