tkgally/eex-dict

Python

0

232 commits

updated Oct 4, 2026

See the code

README

TKG English Learner's Dictionary

An original English-English learner's dictionary for intermediate and advanced learners, written by language models under the direction of Tom Gally, a lexicographer, and published as a static website from this repository.

Every entry in this dictionary was drafted by a language model, reviewed by other language models, and adjudicated by a language model session. No human has checked most entries. The owner reads reports and samples entries. Corrections are welcome as issues.

  • Data (everything under entries/, headwords/, schema/, wiki/, journal/, reviews/, experiments/, metrics/): CC0 1.0. Use it for anything, no attribution required.
  • Code (tools/, .github/): MIT.
  • Openness. The repository contains only the project's own work. No frequency list, word list, pronouncing dictionary, WordNet, or Wiktionary data is vendored, copied, or derived from, whatever its licence. Open resources are consulted at run time only to check a fact, outside the repository, and an entry records only the verdict. The register of resources consulted is sources/README.md. Headword selection and frequency bands are the dictionary's own editorial estimates, made from the models' judgment, not from corpus counts.

How it is built

PROJECT.md is the charter. CLAUDE.md holds the rules every session follows; routine-prompt.md is the scheduled run; wiki/ is the project's knowledge base on the framework in framework.md; journal/ holds a plain-English report per session, also published on the site. Entries are validated by tools/validate.py against schema/entry.schema.json and the closed vocabularies in schema/vocabularies.json; the site is generated by tools/build_site.py in GitHub Actions on every merge to main and is never committed.

Site

The site is deployed by GitHub Pages from the pages workflow, live at https://tkgally.github.io/eex-dict/. Until a custom domain is set, it stays at that GitHub Pages URL.

tkgally/eex-dict

Python

0

232 commits

updated Oct 4, 2026

See the code

README

TKG English Learner's Dictionary

An original English-English learner's dictionary for intermediate and advanced learners, written by language models under the direction of Tom Gally, a lexicographer, and published as a static website from this repository.

Every entry in this dictionary was drafted by a language model, reviewed by other language models, and adjudicated by a language model session. No human has checked most entries. The owner reads reports and samples entries. Corrections are welcome as issues.

  • Data (everything under entries/, headwords/, schema/, wiki/, journal/, reviews/, experiments/, metrics/): CC0 1.0. Use it for anything, no attribution required.
  • Code (tools/, .github/): MIT.
  • Openness. The repository contains only the project's own work. No frequency list, word list, pronouncing dictionary, WordNet, or Wiktionary data is vendored, copied, or derived from, whatever its licence. Open resources are consulted at run time only to check a fact, outside the repository, and an entry records only the verdict. The register of resources consulted is sources/README.md. Headword selection and frequency bands are the dictionary's own editorial estimates, made from the models' judgment, not from corpus counts.

How it is built

PROJECT.md is the charter. CLAUDE.md holds the rules every session follows; routine-prompt.md is the scheduled run; wiki/ is the project's knowledge base on the framework in framework.md; journal/ holds a plain-English report per session, also published on the site. Entries are validated by tools/validate.py against schema/entry.schema.json and the closed vocabularies in schema/vocabularies.json; the site is generated by tools/build_site.py in GitHub Actions on every merge to main and is never committed.

Site

The site is deployed by GitHub Pages from the pages workflow, live at https://tkgally.github.io/eex-dict/. Until a custom domain is set, it stays at that GitHub Pages URL.

Languages

Python

93.6%

JavaScript

3.8%

CSS

2.3%