FAQ

Can I use these commercially?

Yes. MIT, code and icons alike. No attribution, no conditions, forever. That is stated plainly because every extra condition is a reason for someone's legal team to pick a different set — and that decision happens without us ever hearing about it.

Why not just contribute to Lucide?

Five hundred domain-specific AI icons is scope creep for a generalist set, and it should be. IconMind is built to sit next to those sets, not replace them — the 24px grid and 2px stroke match deliberately so both can appear in one interface without clashing.

Icons drawn with AI — isn't that slop?

Every icon passes an automated validator and a human review, and roughly seven of eight candidates are thrown away. The rules and the validator are in the repository; you can read exactly what is enforced.

The honest version: AI moved the bottleneck from drawing to judging. Judging is still the slow part, and it is still done by a person.

Will there be a solid variant?

No. The set's language is open silhouettes — shapes that stop short of closing — and a shape with no inside cannot be painted into anything but a band as wide as the stroke that drew it. Three separate engines were built to derive one, and all three produced cells that either read identically to the outline or came out broken. Duotone carries the weight that a solid variant would have carried.

What if the project stops?

MIT, no backend, no accounts, no telemetry. Whatever you have installed keeps working forever, and the repository can be forked by anyone.

Why is the set so small?

Because the alternative is worse. A thousand mediocre icons is not more useful than a hundred good ones — it is harder to search and harder to trust. The set grows at the rate a person can actually review it.

An icon I need is missing

Request it. Requests and searches that return nothing are what decide the next batch.