Company · About us

The company behind the marketplace

SkillKit is an independent, creator-funded marketplace company. This page collects the facts about us — who we are, how the business works, where we have been, and how to reach the team.

undefined

Operating principles in practice

Principles are cheap until they cost something, so here is what ours look like when they do. "Show the work" has cost us listings: sellers have declined to join because their prompts would be visible, and we accepted the smaller catalog rather than a hidden-content one. "Speed is respect" has cost us features: several shipping ideas were cut because they required account upgrades, checkout redirects or email-confirmation loops that added minutes to flows that take seconds today. "Curate, do not hoard" costs growth every week in the most literal way — every rejected submission is potential GMV we chose to decline.

The compensating reward is compounding trust, which shows up in the numbers that matter to a marketplace: repeat purchase rates, creator retention, and the share of new listings that arrive from creators who first joined as satisfied buyers. Those are the metrics on the wall, and they are why the expensive principles survive budget meetings.

Our stance on responsible AI

SkillKit distributes instructions for AI agents, which makes us a small but real part of the AI supply chain — and we treat that responsibility specifically. Every listing is human-reviewed before it goes live, for what it does and how it instructs agents to behave. Content that is illegal, hateful, or that sexualizes minors is rejected and its submitter removed — no warnings, no second chances. Prompt-injection payloads designed to attack other systems or exfiltrate data are treated as malware. And we publish what we review, so buyers can read a Skill's actual instructions rather than trusting our judgment blindly — transparency is also a safety mechanism, not just a sales one.

We do not overclaim beyond that. We are not a model provider and cannot control what an agent does with a Skill; the agent runtimes' own safety systems are the outer layer, and honest listing descriptions are ours. Where the two layers disagree about a listing, the listing comes down until resolved.

Accessibility

The interface targets WCAG AA contrast on every theme — which is why theme work is measured, not eyeballed — full keyboard operability for navigation, dialogs and forms, semantic landmarks and headings on every page, and visible focus states throughout. Accessibility regressions are treated as bugs with the same 24-hour first-response expectation as anything else. Long-form pages use plain language by policy: if a sentence exists to sound impressive rather than to inform, it gets cut in revision.

Infrastructure & data practices

The product runs on a lean, boring stack — boring on purpose, because marketplace uptime is a trust product. Traffic is served over TLS, financial state lives in an append-friendly ledger design that makes every transaction reconstructable, and personal data practices follow the commitments in the Privacy Policy: no data sales, no advertising trackers, aggregate-only analytics. When something breaks, the incident note goes into the site changelog with the same candor as feature notes — the company that publishes its response targets can hardly hide its outages.