withmarbleapp/os-taxonomy

Marble Skill Taxonomy is a dataset, not an application, that maps out what children learn during the primary and elementary school years. It breaks learning down into 1,590 small teachable ideas called micro topics, such as building sentences or understanding the apparent brightness of stars, and connects them with 3,221 prerequisite links that show what a learner needs to know before moving on to the next idea. Together these form a directed graph rather than a flat list of standards. Each micro topic includes a plain language description, criteria for how you would know a learner has mastered it, a type such as conceptual or procedural, a subject and domain, and an approximate age range. Each prerequisite link is marked as either a hard or soft dependency and comes with a one line reason explaining the connection. The topics are also mapped back to real curriculum standards, including NGSS, Common Core, and the UK National Curriculum, and grouped into 183 parent friendly summaries covering each subject, domain, and age range. Subjects covered include science, mathematics, English, history, personal and social development, life skills, computing, and learning to learn. All the data lives as plain JSON files, with a schema folder describing their structure and a manifest file listing counts and checksums for each file. There is no runtime or dependencies needed to use it: you simply load the JSON files in whatever language or tool you are working with. A small validation script written for Node.js checks that the structure and the internal references between topics are consistent. Licensing here is split into layers. The database itself, meaning the structure, identifiers, and relationships between topics, is under the Open Database License, which allows commercial use but requires attribution and that any modified version of the database itself stay open. The written text Marble created, like descriptions and explanations, is under Creative Commons Attribution ShareAlike. The curriculum standards data was extracted from outside sources and keeps its own original licenses, described separately. Building a product on top of this taxonomy does not require you to open source that product, only improvements made to the taxonomy itself.