Verified creators
Trust in a skill depends on trust in its author. SkillOx runs a five-level identity verification ladder, from pseudonymous-on-GitHub up to legally-verified organization. Buyers can filter the catalog by minimum verification level; the Pro/Team/Enterprise tiers can require it via policy.
The five levels
GitHub account only. Shown with a yellow shield in the catalog. Default level for any new submission.
Phone + email + 2FA confirmed against the creator portal.
Government-ID check via Persona or Onfido. Proves a real person, not a bot.
ORCID, LinkedIn, or professional accreditation (e.g. CPA, CISSP) confirmed. The level required to apply to the Expert Review Network.
Company ownership confirmed via D-U-N-S, EIN, or equivalent legal-entity registry. The label you see on @stripe/, @anthropic/, @hetzner/ in the catalog preview.
What each level unlocks
- L1+ — submit free skills to the public catalog
- L2+ — list paid skills, take payouts via Stripe Connect
- L3+ — apply to the Expert Review Network (planned beta, later public)
- L4 — display the verified-org badge; required for procurement-grade signed releases (Enterprise tier, planned)
Why tiered rather than binary?
Most skill ecosystems are flat — anyone with a registered account can publish, and the only signal is install count. The flat model is what got us to a 401K-skill firehose where 80% is AI-generated slop and 13.4% has critical issues.
The tiered model lets a single, hand-vetted L4 organization publishing a hardened Stripe-integration skill outrank a fresh-account L0 with 10x the downloads. Verification is a slower, more expensive trust signal than stars — that's the point.
Anti-fraud + slashing
Real money + status signals attract fraud. The countermeasures include shadow reviews, ML collusion detection on review-pattern clustering, and reputation slashing: false-negative endorsements that turn out to be malicious drop your rep score by 20 points; three incidents = removal from the verified pool.