← The AI Hype Audit — all 211 verdicts
PARTLY “10 GitHub repos that will change your life” — mostly real, a couple of star counts padded
The claimTen repos with big star counts: mattpocock/skills (203k), anomalyco/opencode (193k, “Claude Code rival”), yt-dlp, garrytan/gstack (126k), unclecode/crawl4ai (76.6k), jamiepine/voicebox (49k), heygen-com/hyperframes, impeccable, microsoft/SkillOpt, trailofbits/skills (“40+ security skills”).
The repos are overwhelmingly real; the inflation is modest, not fabricated. crawl4ai is real but sits around ~60k, not 76.6k. opencode is real and genuinely a Claude Code rival, but ~161–180k, not 193k — and the “anomalyco/opencode” owner is defensible (the project began as sst/opencode and migrated). voicebox (jamiepine, 7 local TTS engines), mattpocock/skills, yt-dlp, garrytan/gstack (Garry Tan’s 23-tool setup, verbatim), heygen-com/hyperframes, and trailofbits/skills all exist as described. Honest corrections: knock crawl4ai to ~60k and opencode to ~170k, and trailofbits shows roughly a dozen-plus skills, so “40+” may be padded. Good list, a few numbers rounded up.
What holds up
- Most repos are real and accurately described — opencode, voicebox, mattpocock/skills, gstack, yt-dlp, hyperframes, trailofbits/skills
- opencode really is a genuine Claude Code rival; the migrated owner name is defensible
- voicebox’s “7 TTS engines, local” claim is exactly right
What doesn’t
- crawl4ai inflated — ~60k, not 76.6k
- opencode inflated — ~161–180k, not 193k
- trailofbits “40+ security skills” looks padded (repo shows ~a dozen-plus)
- impeccable and microsoft/SkillOpt not independently confirmed this pass
The catch
The repos are real — the star counts are just rounded up for the thumbnail. Trust the tools, not the numbers.
How to actually do it
- Explore the repos directly on GitHub — they’re real and mostly free
- Ignore the exact star counts; a couple are inflated by 15–30%
- Match the tool to your actual stack rather than chasing the biggest number
Real repos with a few star counts padded 15–30% — solid list, inflated numbers.
- Confidence
- Medium
- Posted by
- @melibuilds.io (Zach Melinger)
We test hype for free. We build the real thing for a living.
Thirty minutes, no pitch — and you'll leave with something useful either way.
Book a call with Todd or start with the free Business Checkup →The Verdict Weekly
Three verdicts every Friday. Free forever, unsubscribe anytime, no spam — that would be ironic.
© Schreier Group · schreiergroup.com · See a wild AI claim? Drop it here and we'll test it.