Content Source Ledger

UI audio, digital audio, and impact sound packs, plus game art.

Licence

CC0source

What we use it for

Interaction sounds and feedback audio. No attribution required.

Modular, hand-drawn, kid-friendly character illustrations.

Licence

CC0source

What we use it for

Character art in scenes and worksheets. The cleanest licence of any illustration set.

Notes

Preferred over unDraw and DrawKit, which both ban AI/ML training, and over Storyset, which requires visible attribution on the free tier.

Full K-12 scope and sequence, lessons, practice problems, and cool-downs, with the widest grade coverage available under CC BY.

Licence

CC BY 4.0source

What we use it for

K-12 progression reference and problem structure.

Exclusions and gotchas

IM registered trademarks are excluded from the licence — strip all IM branding.

Required attribution

Based on IM® K-12 Math by Illustrative Mathematics®, licensed CC BY 4.0.

Notes

VERSION TRAP: the 2024 v.360 edition is CC BY-NC. Pin the first-edition vintage on every ingested artifact or the provenance cannot be defended later.

The IM 6-8 curriculum in a directly usable package, student and teacher materials.

Licence

CC BY 4.0source

What we use it for

Grade 6-8 sequencing, with all GeoGebra-generated images scrubbed and regenerated.

Exclusions and gotchas

OUR and IM names, logos, covers, and iconography excluded. Images created with GeoGebra carry GeoGebra's own non-commercial terms. Spanish-language-text images are not CC BY.

Required attribution

Adapted from Open Up Resources 6-8 Math, licensed CC BY 4.0. Download for free at openupresources.org.

Notes

The 3rd edition, the K-5 materials, and all high-school materials are CC BY-NC — different licence, same brand and URL family.

Enormous drill volume, worked methods, and the classical mental-arithmetic progressions.

Licence

Public domain (US, published before 1931)source

What we use it for

Classical-track problem structure and mental arithmetic ladders.

Exclusions and gotchas

Project Gutenberg is a registered trademark and may not be used if you charge for the ebooks — strip the PG header, footer, and every PG reference, after which the text is unrestricted.

Notes

Requires OCR cleanup, modernisation of archaic units (rods, gills), and replacement of dated contexts.

Complete Grade 4-12 textbooks with worked examples and large exercise banks.

Licence

CC BY 3.0 Unportedsource

What we use it for

Secondary exercise banks and worked-example structure, re-sequenced from CAPS to US grades.

Exclusions and gotchas

Only the UNBRANDED versions are CC BY 3.0; the branded PDFs are CC BY-ND and cannot be adapted. IT/CAT titles are closed copyright.

Required attribution

Adapted from Everything Maths by Siyavula, licensed CC BY 3.0.

Notes

Underused in the US market precisely because it is CAPS-aligned; the re-sequencing cost is real but one-time.

A complete, printable, chapter-structured middle-school curriculum. The only full, adaptable, commercially clean middle-school curriculum found.

Licence

CC BY 4.0source

What we use it for

Grade 7-8 topic sequencing and problem structure. Start here.

Exclusions and gotchas

University of Utah trademarks are not licensed.

Required attribution

Adapted from the Utah Middle School Math Project, © The University of Utah, licensed CC BY 4.0.

Notes

Licence text: released under CC BY 4.0 allowing free sharing, adaptation, and redistribution for any reason with proper attribution. The project site was unreachable at research time; the licence was verified via a third-party copyright notice and PDFs hosted at uen.org.

State-authored math OER, instructional routines, and rubrics.

Licence

CC BY 4.0 for OSPI-authored contentsource

What we use it for

Routines and rubrics. Small volume but clean and growing.

Exclusions and gotchas

The hub also aggregates third-party items under mixed licences — check per item, never bulk ingest.

100k algebraic word problems with rationales, plus 37k annotated with formal operation programs useful for step verification.

Licence

Apache 2.0source

What we use it for

Operation-program structure for step decomposition.

Notes

Crowd-sourced and noisy; rationales and some answers are wrong. Filter hard.

8.5k human-written grade-school word problems with natural step-by-step solutions — almost exactly the shape a step-based tutor needs.

Licence

MITsource

What we use it for

Seed corpus for step-DAG decomposition and word-problem structure extraction.

12.5k competition problems in LaTeX across 7 subjects and 5 difficulty levels.

Licence

MITsource

What we use it for

High-school and enrichment items.

Notes

Upstream provenance is thin (competition-sourced) even though the release is MIT. Too hard for most K-8.

~3.3k math word problems.

Licence

CC BY 4.0source

What we use it for

Word-problem templates. Small but clean.

395k augmented problems derived only from GSM8K and MATH train splits (no test leakage).

Licence

MITsource

What we use it for

Problem variation patterns.

14M problem-solution pairs built from GSM8K and MATH training splits.

Licence

CC BY 4.0source

What we use it for

Solution-style reference and step-explainer training data. Attribution required.

Required attribution

Includes material from OpenMathInstruct-2 by NVIDIA, licensed CC BY 4.0.

200k grade-school word problems.

Licence

MITsource

What we use it for

Bulk word-problem variation, only after CAS verification.

Notes

Answers are GPT-4-Turbo generated and contain errors. Every item must be independently CAS-verified before it reaches a child.

Kid-readability fonts: Lexend (reduces visual stress), Atkinson Hyperlegible (maximum character distinction — 6/b/9, I/l/1, O/0, which matters enormously in math), Andika (literacy learners), OpenDyslexic (accessibility toggle).

Licence

SIL OFL 1.1source

What we use it for

Lexend as the default UI face, Atkinson Hyperlegible for numerals and expressions.

Exclusions and gotchas

Reserved font names may not be used on derivatives; fonts may be bundled and sold WITH software but not sold by themselves.

Permissively licensed icon sets, no attribution required.

Licence

ISC / MITsource

What we use it for

Lucide is already the project default. Tabler for breadth.

Notes

Avoid Font Awesome Free: it is a three-way licence split (CC BY 4.0 icons / OFL fonts / MIT code) and does require attribution.

Full Foundation-Year 10 mathematics curriculum with elaborations.

Licence

CC BY 4.0 unless indicated otherwisesource

What we use it for

International structural cross-check, filtered for exclusions.

Exclusions and gotchas

Third-party material, logos, photos, videos, and teacher-support resources are excluded; the National Literacy Learning Progressions are CC BY-NC.

Machine-readable JSON of standards for all 50 states plus organisations, with a per-record licence field.

Licence

CC BY 3.0 (via ASN)source

What we use it for

The standards-alignment spine for all 50 states. Highest effort-to-value ratio on the list.

Exclusions and gotchas

ASN must be identified as the source visibly to end users.

KS1-KS4 programmes of study.

Licence

Open Government Licence v3.0source

What we use it for

UK crosswalk, and a future UK market version. The most permissive national standards document mapped.

Required attribution

Contains public sector information licensed under the Open Government Licence v3.0.

GENERATOR CODE for roughly 2M synthetic items across algebra, arithmetic, comparison, measurement, numbers, polynomials, and probability.

Licence

Apache 2.0source

What we use it for

Unlimited provenance-clean drill items. Needs a presentation layer; the phrasing is terse.

Notes

The single cleanest source of practice volume available: generated items cannot collide with anyone's copyrighted bank.

Math rendering (KaTeX/MathJax), student math input (MathLive), interactive geometry and plotting (JSXGraph MIT arm, Mafs), and hand-drawn diagrams (Excalidraw).

Licence

MIT / Apache 2.0 / dual LGPL-MITsource

What we use it for

KaTeX as default renderer, MathLive for equation entry, Mafs and JSXGraph for interactive visuals.

Notes

Since GeoGebra, Polypad, Didax, and Toy Theater are all closed to commercial use, our own manipulatives must be built on Mafs and JSXGraph. Manipulative concepts are unprotectable pedagogy; only specific implementations are protected.

Computer algebra system for independently verifying every generated answer and intermediate step.

Licence

BSD-3-Clausesource

What we use it for

The verification layer. Non-negotiable for a K-12 product: shipping a wrong answer to a child is worse than any licensing issue.

Prealgebra and College Algebra course shells.

Licence

CC BY-SAsource

What we use it for

Reference only. ShareAlike would viralise any derivative and is incompatible with proprietary content.

Prealgebra, Algebra & Trigonometry, and other titles as they stood when distributed under CC BY.

Licence

CC BY 4.0 at time of that distribution (current catalog is CC BY-NC-SA 4.0)source

What we use it for

Blocked pending counsel.

Notes

CC licences are irrevocable, so a verifiably archived copy distributed under CC BY 4.0 stays CC BY 4.0 for the duration of copyright. That requires hard provenance — an Internet Archive snapshot showing the CC BY notice on that exact edition, captured and stored. It is a real path and exactly the kind of claim that must survive scrutiny. Get counsel before relying on it.

Large CoT problem corpora.

Licence

Apache 2.0 / ODC-BY as applied by the compilersource

What we use it for

Only the GSM8K/MATH-derived and synthetic subsets. Audit the source column and drop the AMC/AoPS splits.

Exclusions and gotchas

Includes AoPS forum posts, AMC/AIME problems, and Chinese exam papers OCR'd from PDFs. A permissive tag applied downstream cannot grant rights in scraped third-party content.

1k challenge variations of existing word problems.

Licence

MIT (as applied)source

What we use it for

MAWPS-derived items only. Filter by source field.

Exclusions and gotchas

Constructed by varying ASDiv-a (CC BY-NC) and MAWPS (CC BY) problems. The MIT tag cannot launder the NC upstream.

3,000+ released assessment items.

Licence

Unclear; not published in a fetchable formsource

What we use it for

Do not ingest. Item FORMAT may inform our own item design.

Notes

17 USC 105 covers works by federal EMPLOYEES; NAEP items are built by contractors, and the legislative history deliberately avoids prohibiting copyright in contractor works. This is the most commonly asserted and weakest "it is a government work" claim in edtech.

The US alignment spine: standard codes and text.

Licence

Public licence: royalty-free right to copy, publish, distribute, and displaysource

What we use it for

Codes and links as alignment metadata only. Never present an EzSteps paraphrase as the standard text.

Exclusions and gotchas

Does NOT permit revising or recasting the standards, or selling the standards themselves.

Required attribution

© Copyright 2010. National Governors Association Center for Best Practices and Council of Chief State School Officers. All rights reserved.

Japanese curriculum structure and grade placement.

Licence

No open grant foundsource

What we use it for

Reference for the Japanese crosswalk only.

Texas standards as a certified API and CSV.

Licence

TEA copyright; districts and publishers may use for alignment taggingsource

What we use it for

State crosswalk codes. Verbatim bulk republication is not permitted.

Primary and secondary syllabus structure.

Licence

Personal and non-commercial use only, all rights reservedsource

What we use it for

Reference for the Singapore crosswalk only. The bar-model METHOD is an unprotectable idea and may be implemented freely; the syllabus document may not be reproduced.

K-12 concepts and practice.

Licence

CK-12 Curriculum Materials License (formerly CC BY-NC 3.0)source

What we use it for

Screening corpus only.

Notes

ToS §4.6 explicitly bans scraping content for the purpose of training, developing, or improving any machine learning or AI models. This is contract, not just copyright.

Full K-12 curriculum.

Licence

CC BY-NC-SAsource

What we use it for

Screening corpus only. The NC is fatal and the SA would viralise our content. Great Minds' Eureka is separately all-rights-reserved.

Videos and exercises.

Licence

CC BY-NC-SA 4.0source

What we use it for

Screening corpus only.

Integrated HS math and full K-12 curriculum.

Licence

CC BY-NC-SA 4.0 (Fishtank ToS is more restrictive still)source

What we use it for

Screening corpus only.

Notes

MVP sits next to Utah's CC BY materials on the same state OER hubs and is frequently mistaken for CC BY.

CBSE-aligned textbooks.

Licence

All rights reservedsource

What we use it for

Not usable. NCERT has issued public infringement warnings to publishers.

College and HS textbooks.

Licence

CC BY-NC-SA 4.0source

What we use it for

Screening corpus only.

Notes

RED FLAG: "OpenStax is CC BY" was true historically and is repeated everywhere; it is no longer true. Any blog post or LibGuide saying otherwise is stale.

2.3k diverse word problems.

Licence

CC BY-NC 4.0source

What we use it for

Screening corpus only.

Notes

RED FLAG: routinely bundled with MIT-licensed datasets in "open math datasets" lists. It is not commercially usable, and it contaminates SVAMP.

Competition problems.

Licence

All rights reservedsource

What we use it for

Screening corpus only.

Notes

Sites hosting these do so under negotiated permission, not a public licence. These problems leak into NuminaMath and MATH — a laundering path not to be relied on.

Rich investigations, open-middle problems, formative assessment lessons.

Licence

Non-commercial (MARS is CC BY-NC-ND — NC and ND)source

What we use it for

Screening corpus only. The problem GENRES (open middle, which-one-doesn't-belong, notice-and-wonder) are unprotectable ideas and we may author our own items in those formats — but "Open Middle" is a registered trademark and the name may not be used.

Released assessment items.

Licence

Non-commercial / personal use only, per statesource

What we use it for

Screening corpus only.

Notes

State works get no 17 USC 105 benefit at all, and every release checked asserts non-commercial or personal-use-only terms.

Interactive manipulatives and dynamic geometry.

Licence

Non-commercial; commercial use requires a signed agreementsource

What we use it for

Not usable. Build our own manipulatives on Mafs and JSXGraph instead.

Notes

Mathigon's npm code libraries (@mathigon/core, fermat, euclid, boost) ARE separately open and usable; the textbook content repo says all rights reserved. GeoGebra's terms also contaminate GeoGebra-produced images embedded inside otherwise-CC BY OER, which is why OUR images must be scrubbed.