Skip to content

Latest commit

 

History

History
31 lines (21 loc) · 6.33 KB

File metadata and controls

31 lines (21 loc) · 6.33 KB

Card data

Licensing note: card names, rules text, and game mechanics in this file are Riot Games' copyrighted content, not this repository's. This repo's MIT license (see ../../LICENSE) covers the skill and tooling around this data, not the card content itself. Deck builders and card libraries are among Riot's Developer Portal policy approved use cases in principle, but that approval is conditioned on using assets sourced from Riot's own API — this file is harvested from api.riftcodex.com instead, an unaffiliated fan project, so treat this dataset's standing under the policy as an open compliance question, not a settled one. See the root README.md for the full picture, including this project's pending (not yet approved) product registration.

riftcodex_cards_raw.json — full official English rules text for Riftbound cards, harvested from api.riftcodex.com (an unauthenticated public API run by an unaffiliated fan project — "RIFTCODEX — An unofficial fan project. Not affiliated with Riot Games.").

This is the riftbound skill's own bundled local dataset — read it directly rather than fetching per card question (see ../SKILL.md's "route, don't snapshot" section).

What's in it

  • 1,451 rows as of the 2026-08-16 harvest — that's rows/printings, not distinct cards: 1,304 unique riftbound_ids, with 147 of those ids appearing more than once across the 1,451 rows (294 rows total) as separate reprint/variant printings of the same card (Metal, Overnumbered, Alternate Art, etc.), each with its own row. See extract_legend_packets.py's is_reprint_variant() for how this repo's own tooling filters those back down to one canonical row per real card.
  • Coverage: OGN (Origins), OGS (Proving Grounds), SFD (Spiritforged), UNL (Unleashed), VEN (Vendetta), plus promo pools OPP (Organized Play, 133 cards), PR (general Promo, 13), JDG (Judge Promo, 3).
  • Not covered: RAD (Radiance, unreleased as of this harvest), ARC and FND (riftcodex doesn't carry these at all).
  • Known per-card text gaps (upstream riftcodex data quality, not this repo's processing — found in passing, not from a systematic scan, so assume there are others): Last Rites (SFD-150) — local text.plain carries only the Equip cost line ("[Equip] — Chaos rune, Recycle 2 cards from your trash"); the printed card also has +2 Might and "When I conquer or hold, you may play a unit from your trash" (per wiki.leagueoflegends.com and riftmana.com, checked 2026-08-18). Before quoting any card's full effect as authoritative, treat a suspiciously short text.plain as a signal to cross-check a second source, not as the whole card. Also note name-casing quirks when matching by name: e.g. Ride The Wind (capital T) — an exact-match lookup for "Ride the Wind" misses it.
  • Each card has text.plain and text.rich (HTML) rules text using :rb_*: icon tokens (e.g. :rb_energy_1:, :rb_rune_fury:) for keyword/mana symbols — these are riftcodex's own placeholder syntax, not a standard.
  • No ban-status field. The ban list is a separate, Riot-official, frequently-updated axis — always cross-check the live Rules Hub (https://playriftbound.com/en-us/rules-hub/) or a deployment's private companion binding, never assume this file's absence of a ban field means a card is legal.
  • ID shape: riftbound_id is lowercase with a total-set-size suffix (e.g. ogn-001-298) and */letter suffixes for alternate-art variants (e.g. unl-229*-219) — not the same shape as other Riftbound tools' OGN-001-style IDs. Match on set.set_id (uppercase) + zero-padded collector_number if cross-referencing against another dataset.

Also in this folder

  • errata_overlay.json — official Riot card errata as a corrections layer over the raw file (see ../SKILL.md, "route, don't snapshot").
  • rules_source_registry.json — authority, locale, region, document class, currentness, and supersession metadata for Rule Consult sources.
  • rules_manifest.json — grouped source pointers for local-only English and optional Simplified Chinese rules/FAQ/errata documents. Run ../scripts/bootstrap_rules.py --yes for the controlling English pair, add --include-supplemental-en for English errata plus the historical Origins FAQ snapshot, or add --include-zh-cn; downloaded files and the generated search index remain ignored.
  • rules_query_aliases.json — small bilingual retrieval aliases used by ../scripts/rules_index.py; it improves discovery but never determines a ruling.
  • tournament_lists/ — a small, per-Legend, per-environment store of individually-sourced public tournament decklists (cap 8 per Legend per environment; two environments only; no rates or rankings ever derived from it — the compliance rules are in tournament_lists/README.md and enforced by scripts/check_tournament_lists.py). Exists so Tier 2 rows and deck primers can cite specific real lists instead of re-scraping.

Regenerating

This file is a point-in-time snapshot (2026-08-16), not a reproducible build output — a clean clone of this repo cannot regenerate it. The harvest script that produced it isn't part of this public repo — it lives in the maintainer's own site tooling, unrelated to any card data specific to that deployment, and there's no commitment that it will be published. What is documented here is the manual method for anyone who wants their own independent copy: page through api.riftcodex.com's public card endpoint (unauthenticated, ~15 pages at 100/page as of this harvest, gentle rate limiting recommended since there's no documented limit) and write the result to riftcodex_cards_raw.json in this folder in the same shape. Running that by hand will not reproduce this exact file byte-for-byte (the upstream API's own contents can change between harvests) — it gets you a fresh, independently-sourced snapshot in the same format, not a rebuild of this one. scripts/extract_legend_packets.py is the part of the downstream pipeline that is in this repo — it consumes this file's format directly, whichever snapshot is dropped in.

See the root README.md's Takedown note: this dataset is bundled on the strength of the compliance picture described there, not as a cleared asset, and comes down on request from the rights holder.