Static template — Save to persist, or Regenerate once the LLM client lands.
Problem
Users following forrestchang/andrej-karpathy-skills hit friction the moment they try to apply the trend to their own workflow. AI coding tools attract stars quickly, but maintainers cannot see where new users fail activation or contributor setup.
Audience
Indie hackers + small dev teams already tracking forrestchang/andrej-karpathy-skills. They have intent but not the patience to plumb config + auth + storage.
Outcome
Bring time-to-first-value from hours to under five minutes with sensible defaults and a one-click deploy path.
MVP scope
1) one-click deploy, 2) opinionated config schema, 3) example dataset/seed, 4) a 90-second demo + docs.
Tech stack
Next.js + TS, HOSTUP or self-hosted runtime, Postgres for state, Clerk for auth. Reuse forrestchang/andrej-karpathy-skills as a dependency, no fork.
Distribution
Post on the forrestchang/andrej-karpathy-skills Discord/GitHub Discussions, Hacker News Show HN, plus a Twitter/X thread tagged with the upstream handle.
Risks
Upstream breaking changes; license edge cases; users wanting features that conflict with the simplicity of the starter.
Success signals
≥ 50 deploys in week 1, ≥ 10 GitHub stars on the starter repo, ≥ 1 upstream PR linking back to it.