Slop Stopper Case Study
My notes on this project and why its going on the shelf
Background
Slop Stopper is a Github App that uses heuristic and deterministic code to scan merge requests. It leaves a 'slop score' based on how well scoped the merge is, size of blast radius, and other common code smells.
With the rise of AI agents spewing out unbelieveable amounts of code, a tool that helps reviewers get a quick high level view seems like a good idea. Something like Code Rabbit, but no AI, just a complex but deterministic pattern matching algorithm to keep costs low and speed high.
Building
This project is built with next.js react, hosted on Vercel like most of my projects. Its actually part of the same repo as the blog you're reading now, my portfolio repo ZacDevStory. I wanted to avoid buying another domain until I could get the MVP ready.
(My previous project, yourrelationshipdiagnosis.com, I bought a domain and set up a domain email with Google workspace but it was overkill and a time waster)
I spend days planning, reading Github docs, and implementing the webhook service integration between my API and Github. I was able to get my app working with my account, scanning PRs and leaving a comment with an opinionated score and summary for the reviewers.
Finally, I was ready to submit it to Github Marketplace to get it live and publically available, initally as a completely free alpha testing phase to get user feedback.
Sadly, Github responded 2 days later with a rejection email. No reason given. I responded to the email asking why, but they ghosted me.
I tried emailing support again, but they never responded to my inquiry. The project is effectively DOA, and exists only on my personal Github account as a private and not publically available app.
Summary/Takeaways
I suppose I could have more persistence, more grit. I can try deleting the app and creating a new app from the same repo and re-apply to get the app listed publically.
I could also try self hosting though some other technique, but I would need to redesign how the webhooks and auth function. Considering that, I decided to go back to the drawing board and brainstorm new ideas.
In the future I should really try and avoid reliance on large tech players since they can shut you down with no recourse. Companies like Microsoft are too powerful and a massive threat to autonomy and individual freedom (but that's a different discussion).