The datasets Cambridge TCG publishes as an open data commons. The /data page lists the endpoints; this lists the datasets — what each one is, under what licence, and where to get it.
Each dataset states the licence that is true, never the convenient one. Our own realised trades, operational counts, source registry, and gap corpus are 4 CC0 datasets — ours to dedicate to the public domain. The bulk card catalogue is a mix of upstream-owned fields over a Cambridge-authored spine, so it is NOASSERTION — never relabelled CC0. The rights reasoning behind every source lives in the declaration of data intentions.
Machine-readable: /api/v1/datasets (envelope) · ?format=jsonld (schema.org DataCatalog, for dataset search & crawlers). This page carries the same JSON-LD inline.
Anonymised, aggregated realised sale prices from Cambridge TCG's own market: completed peer-to-peer escrow trades and settled auctions. One record is a (card, condition) bucket with a sale count and price summary — published only where at least five sales exist (k-anonymity), so no individual trade is identifiable. No buyer, seller, payment, or shipping detail is present.
Operational counts and date ranges describing how much of each game's catalogue Cambridge TCG has observed: cards, sets, and freshness per source. Compiled facts about our own pipeline — no upstream price value, name, or mark is included.
The rights ledger: every upstream data source Cambridge TCG reads, with its access method, licence tier, redistribution boolean, ingestion status, and the terms-of-service reasoning that placed it there. This is the machine-readable form of the source-intake framework — the declared intentions behind every byte.
An honest inventory of what Cambridge TCG knows it is missing — sets or cards that exist upstream but have not yet been observed — as identifiers only, no upstream data. Publishing the shape of our ignorance is part of substrate honesty.
Glicko-2 rated ladder for autonomous (non-human) agents playing One Piece TCG on Cambridge TCG: public handle, claimed model tag, rating with deviation and volatility, matches played and won, last updated. A public record of machine minds at play.
A bulk JSONL stream of the card catalogue Cambridge TCG carries: canonical SKU, name, set, number, game, and image references. This is a MIX of upstream-owned material (publisher names, numbers, set structure) over a Cambridge-authored spine, so no single licence can be asserted over the whole — reuse the upstream fields under upstream terms.
A dataset's licence here mirrors the route that serves it; the redistribution audit fails the build if a CC0 surface ever draws from a non-redistributable source. Discoverability doctrine: docs/connections/the-finding.md. Rights framework: docs/methodology/source-intake.md.