Marble Skill Taxonomy is an open, structured dataset of learning outcomes across primary and elementary education, decomposed into fine-grained micro-topics and organized as a prerequisite graph.
The taxonomy solves the problem of fragmented curriculum data by providing a connected graph of learning rather than flat lists of standards. It contains 1,590 micro-topics—each representing a single teachable idea with plain-language descriptions, mastery evidence criteria, subject classification, and age range—linked by 3,221 prerequisite dependencies tagged as hard or soft with explanatory reasons. Each micro-topic is aligned to national curriculum standards including NGSS, Common Core, and the UK National Curriculum. The data is distributed as JSON files with accompanying JSON Schemas for validation, requiring no runtime or external dependencies to use.
Adoption suits teams building educational software, curriculum planning tools, or learning platforms that need structured, standards-aligned learning progressions. The taxonomy covers eight subjects with particular depth in science and mathematics, and includes parent-friendly domain summaries for each subject-age combination. The dataset's prerequisite graph enables features like prerequisite tracing—showing learners everything they must master before tackling a concept—and dependency visualization. Anyone evaluating this should note it is pure data with no accompanying runtime; integration requires loading and querying the JSON directly.
The project maintains a stable identifier scheme for topics and dependencies, enabling reliable cross-referencing. Validation tooling is provided to check structural integrity and referential consistency across the dataset files. The taxonomy includes a manifest file documenting counts, per-subject breakdowns, and SHA-256 checksums for each data file.
