Marble Skill Taxonomy

Overview

Marble Skill Taxonomy is an open, structured taxonomy of what children learn across the primary/elementary years. It decomposes learning into 1,590 fine-grained micro-topics across eight subjects, wires them into a directed acyclic prerequisite graph with 3,221 edges, aligns topics to national curriculum standards, and includes parent-friendly domain summaries.

Key Features

• 1,590 micro-topics, each with a plain-language description, mastery evidence criteria, type, subject and domain, and approximate age range.
• 3,221 prerequisite dependencies represented as a directed acyclic graph, with edges tagged hard or soft and carrying a one-line reason.
• Curriculum alignment linking micro-topics to standards from NGSS, Common Core, the UK National Curriculum, and more.
• 183 parent-friendly one-paragraph domain summaries per subject, domain, and age band.
• Coverage across eight subjects: Science, Mathematics, English, History, Personal & Social Development, Life Skills, Computing, and Learning to Learn.
• UTF-8 JSON data files, JSON Schemas, and a manifest with counts and SHA-256 checksums.

Use Cases

• AI tutoring and retrieval-augmented generation systems that need structured curriculum knowledge.
• Agent queries about what must be learned before a concept and what paths lead to a learning goal.
• Knowledge-gap diagnosis and personalized learning planning using assessment results and prerequisite relationships.
• Curriculum analysis and exploration through a connected graph of learning topics.

Getting Started

• Load data/topics.json and data/dependencies.json as UTF-8 JSON; no runtime or dependencies are required.
• Build an ID map from topics and filter dependencies by topicId to retrieve prerequisites for a topic.
• Validate structure and referential integrity with node scripts/validate.mjs.

Deployment & Requirements

• No runtime or dependencies are required; the dataset is pure UTF-8 JSON.
• JSON Schemas are provided in schema/, and data/manifest.json provides counts and SHA-256 checksums.

Before You Adopt

• License: ODbL-1.0. Review its terms before using, modifying, or distributing the project.
• Curriculum standards are extracted from third-party frameworks, so users should account for the external origin of those standards.
• The release deliberately omits semantic embeddings and any per-child or user data.