DeepSeek Harness Plugin

xczhanjun/lazeword

Stars ★ 0 Category Just for Fun Added 2026-08-15

A cozy offline vocabulary trainer for the whole family: 1094 curated words, spaced repetition, 6 quiz types, spelling games and a lazy large-screen mode; opens as a sidebar panel and also runs standalone as one HTML file.

Install

# from GitHub (first run asks for allowBuilds approval — follow the hint, retry)

dsh plugin --profile web add github:xczhanjun/lazeword

Any plugin you install runs third-party code with your own permissions — it can read your files, use your credentials, and reach the network, and tool approvals don’t sandbox it. GitHub-sourced plugins also run build scripts at install time. Only install sources you trust, and pin a commit (github:owner/repo#sha).

README

中文

A cozy, offline vocabulary trainer for DeepSeek Harness (dsh). Learn 1000+ of the most common English words (plus Hong Kong junior-secondary math & science terms) — with spaced repetition, quizzes, spelling games, reference grammar, and a lazy large-screen "lie-down" mode. Works as a dsh web plugin and as a standalone single-file app.

Features

  • 1094 curated words — 947 everyday words + 147 HK math/science terms (English ↔ 中文/繁體)
  • Spaced repetition — Ebbinghaus intervals (1→3→7→15→30 days) + mistake notebook
  • 6 quiz types — meaning, word, IPA, listening, spelling (letter grid), cloze (fill-in-the-blank from real example sentences)
  • Pronunciation — TTS read-aloud, IPA syllable & stress highlighting that follows the speech, repeat-after-me scoring
  • Word details — example sentences, synonyms, etymology, root/affix hints (offline)
  • Reference — 111 irregular verbs, 24 grammar points, 43 phrasal verbs, 50 conversational sentences
  • Games — space word-matching shooter (click-to-match, speed control, sound effects)
  • Family features — up to 4 learner profiles, parent/teacher report, daily 10 words, JSON backup, study streaks
  • Lazy mode — dark comfortable theme, big fonts, auto-pronounce + auto page-turn, voice control ("会 / 不会"), Space to pause
  • Fully offline — one HTML file, zero dependencies (online only for example sentences / dictionary audio)

Screenshots

Running inside DeepSeek Harness (real installation test):

lazeword in dsh sidebar

The panel opens the full app (1094 words, fully offline):

lazeword panel in dsh

Install as a dsh plugin

dsh plugin --profile web add dsh-lazeword

Then look for 🛋️ 躺着背单词 in the sidebar. The panel opens the standalone app inside the harness (blob-URL iframe — no server, no network, deterministic).

Run standalone

Open app/lazeword.html directly in any browser (double-click).

standalone app Or serve it for phones on the same LAN:

python3 -m http.server 8000
# open http://<your-ip>:8000/app/lazeword.html

Develop

Requires Node ≥ 22.19. Zero npm dependencies (tests use the built-in node:test).

npm test                          # unit tests (core functions)
node scripts/build.mjs             # build the standalone app
node scripts/build-client.mjs      # embed the app into the dsh client bundle
npm run check                     # syntax-check plugin entrypoints

Architecture

src/core.mjs        deterministic pure core (single source of truth):
                    SRS scheduling, IPA syllable/stress parsing,
                    orthographic syllabification, grading, seeded RNG,
                    event log (append-only learning trajectory)
app/template.html   the app UI shell (__CORE__ / __WORDS__ markers)
data/*.json         word lists (Vocabineer 947 + HK subjects)
lib/index.js        dsh host entry (plugin registration)
lib/client.js       dsh web client bundle (embeds the app; generated)
scripts/            build + extraction tooling
tests/              node:test unit tests (20 assertions, zero deps)

Design principles: deterministic, offline, auditable — the app is statically composed at build time (no runtime plugin loading), and the core event log makes the whole learning trajectory reproducible ("spatiotemporal determinism").

License

MIT

Disclaimer

DeepSeek Harness is in developer preview and may introduce breaking changes. This plugin is community-maintained and not affiliated with DeepSeek.

Content from the project README on GitHub ↗

Links

More in this category

View the whole category →