stackmaven is the publication brand i’ve been pouring evenings into: opinionated dev/ai tool catalog with a news layer and (later) a stack recommender. the site went live a few days ago. today was about locking down what the version number actually means and tightening the hero so it stops feeling like an empty room.
versioning, pegged to the roadmap
the new scheme:
- v0.1: phase 1 shipped (scaffold + seed catalog). that was 5/15
- v0.2: phase 2 (catalog depth + news cadence). where we are now
- v0.3: phase 2 complete (250 entries, sustained news, 1k subs)
- v1.0: the recommender ships and monetization is live
free-floating semver tells you nothing. pegging each version bump to a phase in the roadmap means every future bump has an unambiguous meaning. “v1” gets reserved for the thing that actually makes stackmaven stackmaven (the recommender), not just “the site went live.”
the in-product signal is a small v0.2 · alpha chip next to the
wordmark in the header. amber styling, tooltip explains, click-through
to /now for current state. deliberately not a “work in progress”
banner. editorial publications don’t wear hazard tape, and stackmaven
is supposed to read like a publication, not a beta-staged saas app.
fill negative space with your own content
the hero was text-heavy on the left with a dead column on the right at desktop. could’ve reached for a generic illustration. most saas sites do: abstract gradient, swooping lines, the works.
but no editorial publication you respect has a stock graphic in the hero. the new yorker doesn’t. the wirecutter doesn’t. they fill that space with their own content: pull-quotes, picks, recent issues. so that’s what landed: an editor’s picks vertical panel on the right, 11 hand-curated entries sorted by canonical category order so it reads as a tour, not a chronology.
now the headline (“opinionated and current”) gets proven in the same viewport instead of just claimed.
one pattern from the panel i want to keep using: mask-image fade + ghost scrollbar. the list is taller than the panel so you can scroll to see all 11, but the bottom edge fades into the background and the scrollbar is barely visible at rest:
.picks-list {
mask-image: linear-gradient(to bottom, black calc(100% - 40px), transparent);
scrollbar-width: thin;
scrollbar-color: rgba(255, 255, 255, 0.08) transparent;
}
.picks-list::-webkit-scrollbar { width: 4px; }reads as “publication content extending below the fold” rather than
“scrollable list with a clipped edge.” reusable for any bounded
surface that needs a “more below” affordance without a saas-y show more button.
github stars are a directory tic
almost wired up github star counts on every tool card. the cost was nothing. github’s rest api is 5k/hr authenticated, a daily action rewriting frontmatter and letting vercel rebuild is under a minute of ci per day.
reason i held off: stars are a directory tic. futurepedia,
theresanaiforthat, stackshare: every catalog leans on them.
popularity becomes the signal. stackmaven is positioned against
directory framing on purpose: the stackmaven_verdict paragraph on
each entry is the signal. if you want a number, the verdict comes
first.
probably dropping stars from the cards entirely and keeping them on detail pages as one stat among several. context, not headline.
editorial voice is mostly a series of small NOs: no hero illustration, no “work in progress” banner, no github stars on the cards. each one is tiny on its own. the cumulative effect is what makes a publication read like a publication instead of a directory with a fancier homepage.