Sources and copyright
How this service treats other people's work, stated plainly.
What we store
This service converts public trading card game news into structured facts. What is kept is the fact — a set name, a release date, a banned card, a price — together with metadata and a link back to the source.
We do not store or republish full articles. A short excerpt of parsed page text is retained for operator debugging, and individual facts may carry a quoted span of at most 280 characters as evidence for that specific claim. Titles and summaries shown on this site are written by the extraction step, not copied from the publisher.
Every fact links back to the page it came from. If you want the story, read it at the publisher.
How we access sources
- robots.txt is obeyed. If a path is disallowed for our user agent, we do not fetch it. If robots.txt cannot be resolved, we do not crawl.
- An honest user agent identifying the crawler and a contact URL is sent on every request.
- Rate limits are respected. A minimum delay is enforced between requests to the same host, and any crawl-delay in robots.txt overrides it. A 429 backs off; it is never retried around.
- Walls are respected, not bypassed. A 401, 402, 403 or anything that looks like an auth, paywall or bot barrier stops that fetch permanently and is recorded as skipped. This system contains no code for solving CAPTCHAs, rotating identities, defeating bot protection, or accessing content behind a login or paywall — and none should be added.
- Conditional requests (ETag / Last-Modified) are used so unchanged pages are not re-downloaded.
- noindex is honoured. Pages declaring noindex/noarchive are not stored.
Trust weighting
Each source carries an operator-set trust weight and a flag for whether it is a first-party publisher channel. These feed the confidence model: an official announcement outranks any number of outlets repeating each other, and corroboration is counted by distinct publisher rather than by distinct URL.
The weights are editorial judgements made by whoever runs this deployment. They are visible on the sources page and through the API, so you can disagree with them knowingly.
Accuracy
Facts are extracted by a language model and validated against a strict schema before being stored. The model is instructed to return null rather than guess, and anything that fails validation is rejected and surfaced to an operator rather than silently saved.
That is not the same as being right. Treat rumour as unverified, read confidence before relying on something, and follow the source link when it matters.
Trademarks
Pokémon, Magic: The Gathering, Yu-Gi-Oh!, One Piece Card Game and Disney Lorcana, along with all set, card and product names, are trademarks of their respective owners. This service is not affiliated with, endorsed by or sponsored by any of them. Names are used descriptively, to identify what a fact is about.
Removal requests
If you publish a source indexed here and want it removed, the operator of this deployment can disable it in the admin area, which stops all future fetching, and delete its stored records. Contact details are configured per deployment — see the contact URL in the crawler's user agent string.