Datasets:

Wysme
/

riftbound-cards

license: other
license_name: riot-legal-jibber-jabber
license_link: https://www.riotgames.com/en/legal
pretty_name: Riftbound TCG Card Database
tags:
  - riftbound
  - tcg
  - trading-card-game
  - card-game
  - riot-games
  - league-of-legends
size_categories:
  - 1K<n<10K
task_categories:
  - other
configs:
  - config_name: default
    data_files:
      - split: train
        path: cards.csv
  - config_name: errata
    data_files:
      - split: train
        path: card_errata.csv

Riftbound TCG Card Database

Machine-readable snapshot of every card in Riftbound: The League of Legends TCG, auto-scraped from the official Card Gallery and errata pages.

Snapshot date: 2026-08-11 Cards: 1188

Source repo (scraper + pipeline): https://github.com/LouisCourrian/riftbound-cards Every GitHub release publishes the same three formats as attached assets and mirrors them here.

Files

File	What
`cards.csv`	Full corpus, one row per card. Array fields (`card_type_labels`, `tags`) are JSON strings.
`cards.json`	Same corpus, with arrays as real arrays.
`card_errata.csv`	Merged errata (auto-scraped + manual overrides).
`riftbound.db`	Prisma-managed SQLite with a `cards` and a `card_errata` table.

Quick start

Load the primary card list with the datasets library:

from datasets import load_dataset
ds = load_dataset("Wysme/riftbound-cards", split="train")
print(ds[0])
# {'card_code': 'ven-r01', 'name': 'Fury Rune', 'domain': 'Fury', ...}

Load the errata list:

errata = load_dataset("Wysme/riftbound-cards", "errata", split="train")

Query the SQLite directly (nicer for joins):

from huggingface_hub import hf_hub_download
import sqlite3, pandas as pd
path = hf_hub_download("Wysme/riftbound-cards", "riftbound.db", repo_type="dataset")
con = sqlite3.connect(path)
pd.read_sql("SELECT name, domain, energy, might FROM cards WHERE card_type='Unit' AND domain='Calm'", con)

Schema

Notable columns on cards:

card_code — stable identifier (<setCode>-<cardNumber>, e.g. ven-sp2-006)
name / subtitle / full_name — split for champions (Sona / Harmonious / Sona, Harmonious)
card_type — primary type (Unit, Rune, Spell, Gear, Legend, Battlefield, Token)
card_type_labels — every label displayed under Card Type (["Unit","Champion"]); JSON string in the CSV, real array in the JSON
tags — creature type / faction tags; same shape as card_type_labels
energy, power, might — integers or null
domains — every domain symbol printed in the card frame; JSON string in the CSV/SQLite, real array in the JSON. Legends are usually bi-domain (["Fury","Chaos"])
domain — first entry of domains, kept as a plain string for consumers that expect a single value (Calm, Fury, Mind, Body, Chaos, Order, Colorless)
ability_original — the printed text
ability_corrected — the errata replacement (null when none applies)
ability_effective — ability_corrected ?? ability_original (what you should treat as canonical)
has_errata — boolean shortcut
image_url — full-resolution CDN URL (query string stripped)
source_url — deep-link into the Card Gallery

Power's rune colour always equals the card's domain — that's the Riftbound rule, so no separate power_domain column is stored.

Last diff

Total: 1188
New: 9
Changed: 3
Missing (kept in DB, absent from latest scrape): 1

Legal

Card data (names, subtitles, ability text, artwork URLs, set names, artist credits) is © Riot Games. Redistributed here under Riot's Legal Jibber Jabber fan-content policy and the Riftbound TCG Digital Tools Policy for non-commercial community use. Downstream users are responsible for their own compliance with Riot's policy — in particular, do not build gameplay-simulation clones or aggregate metagame data from this dataset.

This dataset is not affiliated with, endorsed by, or sponsored by Riot Games. Riot may revoke the redistribution license at any time; if you're from Riot and want a change, please open an issue on the source repo.

The scraper code itself is MIT-licensed.

Snapshot published on 2026-08-11 by the auto-updater; the source repo is the canonical view.