Credits. Built by
Sebastián R. Vansteenkiste via
DeepSeek V4 Flash @
FREEBUFF (affiliate link) — an independent, from-scratch
reimplementation inspired by tangorin.com
(developed by Gregory Bober, now owned by Archie Preston). omakase is not
affiliated with Tangorin, EDRDG, or any other source or contributor.
Dictionary data. Built from open datasets, each with its own
license and attribution:
- JMdict, KANJIDIC2, kradfile-u, radkfile — © James William
Breen & The Electronic Dictionary Research and Development Group (EDRDG),
CC BY-SA 4.0
(licence statement);
kradfile-u further © Jim Rose (KanjiCafe.com).
- KanjiVG stroke-order diagrams — © Ulrich Apel,
CC BY-SA 3.0
(kanjivg.tagaini.net).
- JmdictFurigana ruby segmentation — © Doublevil,
MIT; data derived from
JMdict (CC BY-SA 4.0).
- jmdict-simplified JSON conversion of the EDRDG files —
© scriptin,
MIT packages; data CC BY-SA 4.0.
- Tatoeba example sentences — © Tatoeba contributors,
CC BY 2.0 FR
(some CC0). Test fixtures only — the shipped dictionary carries no
sentences, so the app never shows one.
- Catalan query index (not dictionary content: it lets a
Catalan word be searched in Spanish, which is the only language the
dictionary glosses it in) — © the Apertium project,
apertium-spa-cat,
GPL-2.0.
Generated from that data, so the table ships under GPL-2.0 while the app
stays MIT; the verbatim licence is in
LICENSES/GPL-2.0.txt.
Software.
better-sqlite3 (MIT) ·
@sqlite.org/sqlite-wasm (Apache-2.0;
SQLite itself is public domain) · dev tooling: TypeScript (Apache-2.0),
tsx (MIT), puppeteer-core (Apache-2.0). Full license texts ship with
the app in LICENSE.md (plus the vendored
GPL-2.0 text); per-component
notices live in NOTICE.md.
Disclaimers. Dictionary data is provided without any warranty as
to its accuracy or suitability (EDRDG licence); it may contain errors, and users
assume all liability for its use. Stroke diagrams come from the bundled KanjiVG
set and are cached locally. The app works 100% offline — nothing you look up
ever leaves this device.