Methodology

How paths are computed

The recipe graph contains millions of cycles, so a naive search never terminates. Each element carries a crafting depth, the shortest chain that can produce it, where the four base elements are depth 1. When solving an element we only recurse into ingredients that are strictly shallower, which guarantees termination, and we explore the cheapest branches first. Paths for the most-searched elements are precomputed; the rest are solved on demand and then cached.

How recipes are verified

Combinations come from the open datasets listed on our data sources page. Each recipe page shows the date its path was last checked against the graph. If you find a combination that no longer works, tell us and we will correct it.

How old URLs are preserved

The previous site used the pattern /how-to-create-…-in-infinite-craft/. Web archives only retained a fraction of those pages, so rather than relying on a fixed redirect list we parse the element name out of any old URL and forward it to the current page. Old links keep working whether or not we ever saw them.

What we do not do

Pages are generated from data, not written by a language model. Elements with no demonstrated search demand stay out of the index until that changes, so we are not publishing hundreds of thousands of near-identical pages for their own sake.