March 25, 2026
Zero stars, zero chill
90% of Claude-linked output going to GitHub repos w <2 stars
AI code is piling up in no‑star repos — and devs are like, “welcome to my world”
TLDR: A new stat claims most Claude‑assisted code ends up in little‑noticed GitHub projects, sparking a fight over whether that’s alarming or just how coding usually looks. Commenters piled in to say most repos are small and unstarred anyway, turning the “90%” headline into a reality check on hype versus popularity metrics.
An eye‑catching claim hit the feed: since launch, 90% of Claude’s AI‑assisted code lands in GitHub projects with fewer than two “stars” (think: likes). One commenter even tallied “50 billion lines” created, with just “5 billion” in projects that have 2+ stars, calling it “eye opening.” Cue the internet debate, sarcasm, and self‑owns in the comments under the HN thread.
On one side: the “AI is flooding GitHub with junk” crowd, clutching the numbers as proof of a noise tsunami. On the other: a chorus of devs saying, basically, this is normal. Most projects are tiny, personal, or experimental — so of course they have zero stars. “Same with human code,” multiple users shrugged, with one betting that way more than 90% ends up in unstarred or private projects anyway.
The mood turned hilariously self‑deprecating. Commenters confessed that 100% of their code lives in ghost towns. Others roasted the metric itself: stars are a popularity contest, not a quality score. Meanwhile, a stray commit note like “🤖 Generated with Claude Code” was held up as the new scarlet letter — or a shiny badge, depending on your vibe. The real drama isn’t the stat; it’s the existential question: is AI filling the world with forgettable code, or are we just seeing the same old developer basement — now with a robot roommate?
Key Points
- •The article states that around 90% of Claude-linked code contributions go to GitHub repositories with fewer than two stars.
- •The analysis focuses on original (non-fork) repositories with their first observed Claude Code commit within the last seven days.
- •An “Activity Over Time” section indicates a temporal analysis of commit activity.
- •A sample commit message shows identification markers: “Generated with Claude Code” and “Co-Authored-By: Claude <noreply@anthropic.com>.”
- •The piece frames findings as observations on momentum, adoption, and clustering of activity by repository star levels.