the five checks, and which one fails the most submissions
we review every submission by hand against the same five checks. here's what each one actually looks for, and why check five throws out more listings than the other four combined.
every submission goes through the same five checks, in the same order, run by the same person. most don't clear all five. here's what each one is actually checking for.
functional
runs from a clean install. we do that install ourselves, not from a screen recording the seller sends us.
source available
repo access on request, with a commit history we can actually read.
proof of substance
sustained use, not one transaction. a small mrr that holds, a user count over time, or a finished audit. self-generated payments do not count.
honest scope
the listing matches the repo, and says plainly what the product does not do.
engineered, not generated
real engineered work, not something prompted into existence. we read the commit history and we can tell.
check five isn't a tiebreaker. it's the reason the other four are worth running at all.
check five throws out more submissions than the other four combined. a repo with a single commit titled 'initial commit' and eleven thousand lines in it does not pass, no matter what the usage numbers say. a project that grew in visible steps, with the mistakes still in the history, usually does.
we don't publish the rejection itself, but we do tell the seller which check it failed on. some resubmit with a real repo. most don't come back, which is roughly the point.
written by solstead team