Morphowiki

/ingest — 分析並收錄單字

這是本專案的核心指令。只要提供一個英文單字，系統便會自動完成語素拆解、建立字典頁面、設置雙向連結並更新索引 —— 所有流程一氣呵成。

/ingest — Analyze and add a word

The workhorse command of this project. You give it an English word; it does the morpheme breakdown, writes the dictionary page, links it both ways to the morpheme files, and updates the index — all in one go.

為什麼需要這個功能？

若要手動將單字加入字典，您通常需要執行以下繁瑣步驟：

  1. 查詢 IPA 發音、撰寫中文釋義並考據詞源。
  2. 將單字拆解為字首、字根或字尾（語素 Morpheme：組成單字的最小意義單位。例如將 unbelievable 拆解為 un-、believe、-able）。
  3. 在 dictionary/words/<word>.md 建立新檔案，並確保 Frontmatter 與標題格式正確。
  4. 分別前往 dictionary/prefix/、dictionary/root/、dictionary/suffix/ 對應的語素檔案，在「Words containing this」清單中手動填入新單字並維持字母排序。
  5. 手動更新 dictionary/index.md 索引。

每個單字至少需調整 4 至 6 個檔案，極易漏掉反向連結。/ingest 透過自動化流程解決了這些問題，讓您只需專注於學習單字本身。

Why this exists

To add one word to this dictionary by hand, you would normally need to:

  1. Look up the IPA, write a definition, dig up the etymology.
  2. Split the word into prefixes, root, and suffixes (morphemes — the smallest meaningful units of a word, e.g. unbelievable splits into un-, believe, -able).
  3. Create a new dictionary/words/<word>.md file with the correct frontmatter and sections.
  4. Open each matching morpheme file under dictionary/prefix/, dictionary/root/, dictionary/suffix/ and add the new word to its "Words containing this" list, in alphabetical order.
  5. Update dictionary/index.md.

That's 4–6 files per word, and it's easy to forget a back-link. /ingest does the whole thing automatically. You only supply the word.

操作範例

在 Claude Code 對話框輸入：

/ingest unbelievable

稍等片刻，Claude 會建立完成的單字檔案。dictionary/words/unbelievable.md 內容範例（節錄）： 

---
word: unbelievable
pos: adj.
ipa: /ˌʌn.bɪˈliː.və.bəl/
added: 2026-05-24
---

# unbelievable

**詞性**: adj.
**IPA**: /ˌʌn.bɪˈliː.və.bəl/
**中文釋義**: 難以置信的;不可思議的

## 拆解
- 字首: [[../prefix/un-]] `un-` — 否定、相反
- 字根: [[../root/believe]] `believe` — 相信
- 字尾: [[../suffix/-able]] `-able` — 形容詞字尾,「可…的」

## 詞源
…(說明 believe 源於古英語 belyfan 等歷史背景)

## 記憶法
…(連結語素或意象的記憶提示)

## 筆記區

在此次操作中，以下三個語素檔案的 ## Words containing this 區塊也會自動填入 - [[../words/unbelievable]]：

  • dictionary/prefix/un-.md
  • dictionary/root/believe.md
  • dictionary/suffix/-able.md

若對應的語素檔案尚不存在，/ingest 會一併建立。這種「雙向連結」確保了單字與語素間的知識網絡是完整且可互查的。詳情請見 什麼是 Morphowiki。

語言設定：上方範例中的釋義與詞源為中文（zh-TW），因為系統預設服務中文使用者。輸出語言可透過專案根目錄的 CLAUDE.md 進行調整（如改為全英文），詳見 自訂與微調。

A concrete example

In the Claude Code chat, type:

/ingest unbelievable

After a few seconds, Claude reports the breakdown and writes the files. The new dictionary/words/unbelievable.md looks roughly like this (abridged): 

---
word: unbelievable
pos: adj.
ipa: /ˌʌn.bɪˈliː.və.bəl/
added: 2026-05-24
---

# unbelievable

**詞性**: adj.
**IPA**: /ˌʌn.bɪˈliː.və.bəl/
**中文釋義**: 難以置信的;不可思議的

## 拆解
- 字首: [[../prefix/un-]] `un-` — negation, reversal
- 字根: [[../root/believe]] `believe` — to trust as true
- 字尾: [[../suffix/-able]] `-able` — adjective suffix, "capable of"

## 詞源
…(two to four sentences on the history of believe, from Old English belyfan, etc.)

## 記憶法
…(one to three sentences of a mnemonic linking the morphemes)

## 筆記區

In the same operation, each of these three morpheme files has a new line - [[../words/unbelievable]] added under its ## Words containing this section:

  • dictionary/prefix/un-.md
  • dictionary/root/believe.md
  • dictionary/suffix/-able.md

If any of those morpheme files don't exist yet, /ingest creates them. This is the "two-way link": the word page points to morphemes, and the morphemes point back at every word that uses them. Read What is a Morphowiki for the full picture.

About the language: the sample above shows Chinese definitions and notes because this dictionary defaults to Chinese-speaking users. The actual output language is controlled by CLAUDE.md in your project root and can be switched to English, Japanese, and so on — see Customize & tune.

About section names: the section headings ## 拆解 / 詞源 / 記憶法 / 筆記區 (Breakdown / Etymology / Mnemonic / Your notes) are fixed in Chinese for now — they are how the review web app finds the content. Don't rename them by hand.

逐步操作步驟

  1. 確認已執行過 /setup（確保 dictionary/ 資料夾與 stopwords.md 已建立）。
  2. 在專案根目錄啟動 Claude Code（在終端機輸入 claude）。
  3. 輸入 /ingest指令並接續您想收錄的英文單字，例如： 
    /ingest predictable
  4. 等待系統分析。成功後，系統會列出拆解詳解（字首/字根/字尾）、新建檔案與更新檔案的清單。
  5. 開 dictionary/words/predictable.md 確認結果。您可以自由編輯 ## 筆記區 以記錄個人例句或學習心得；其餘區塊則建議交由 /ingest 或 /lint 協助管理。

批次收錄與精確過濾：您可以一次輸入多個單字，甚至是一長串句子。/ingest 會自動斷詞、過濾常見虛詞（Stopwords），並為所有新字建立檔案： 

/ingest predictable formation rebuild
/ingest The international cooperation project was a success.

在句子模式下，the、was、a 等單字會被 Stopwords 名單過濾。收錄完成後，系統會提供一份簡要的統計報告。

遞迴式收錄：若收錄的單字包含其他具獨立意義的複合同根字（例如 international 內含 national 與 nation），/ingest 會主動詢問或自動為這些單字建立頁面。單次操作可能會一次產生數個相關單字，這有助於建立完整的字群記憶。

Step by step

  1. Make sure you've already run /setup (it creates dictionary/ and stopwords.md). If not, go do that first.
  2. Open Claude Code from inside the project folder (run claude in your terminal).
  3. In the chat box, type /ingest followed by an English word, for example: 
    /ingest predictable
  4. Wait for Claude's report. It lists the breakdown (prefix / root / suffix), the new files it created, and the existing files it updated.
  5. Open dictionary/words/predictable.md to check the result. You're free to edit the ## 筆記區 ("Your notes") section to add your own example sentences or teaching reminders — leave the other sections to /ingest and /lint.

Batch and whole sentences: you can pass several words at once, or even a full English sentence. /ingest tokenizes it, filters out common function words, and ingests each new word once: 

/ingest predictable formation rebuild
/ingest The international cooperation project was a success.

Recursive intermediate words: if the word you ingest contains shorter English words inside it (e.g. international contains national and nation), /ingest will also create dictionary pages for those intermediates. A single command can produce two or three extra word files. This is intentional.

常見錯誤與下一步

常見錯誤 1 — 收錄太簡單的字:像 the、a、it、is 這類字,被列在 dictionary/stopwords.md 的白名單裡。對它們執行 /ingest 會被拒絕,回應類似「the 已在 stopwords (default),不會建立字典頁」。這是刻意的——這些字不該佔複習庫的位置。要看完整列表,直接打開 dictionary/stopwords.md。

常見錯誤 2 — 拼錯字:/ingest belivable(漏一個 e)還是會被當合法輸入處理,Claude 會努力給出一個拆解。先在 Google 或字典查一下拼寫,再 ingest。

常見錯誤 3 — 已經收錄的字:如果該單字已經有 dictionary/words/<word>.md,/ingest 會回報「已存在」並跳過,不會覆蓋你之前手動加在 ## 筆記區 的內容。這是保護機制。真的要重來,先把舊檔刪掉再 ingest。

常見錯誤 4 — 給中文字:這個 skill 不處理中文。給中文詞它會問你要做什麼。

下一步:

  • 想用現有的字根字首自動造幾個新單字練手?試 /new-word。
  • 想確認整個字典的雙向連結都對得上?跑 /lint。
  • 想把新收錄的字加入你的複習庫?用 /flashcard。
  • 想換輸出語言(中文 → 英文 → 日文…)或微調別的細節?見 自訂與微調。

回 首頁。

Common mistakes & next steps

Mistake 1 — Ingesting a word that's too simple: words like the, a, it, is are in the allow-list at dictionary/stopwords.md. Calling /ingest on them is refused with a message like "the is in stopwords (default), no dictionary page will be created." This is intentional — these words don't belong in a vocab-learning deck. Open dictionary/stopwords.md to see the full list.

Mistake 2 — Misspelling: /ingest belivable (missing an "e") will still be treated as a valid input; Claude will do its best to break it down. Spell-check first.

Mistake 3 — Already-ingested word: if dictionary/words/<word>.md already exists, /ingest reports "already exists" and skips, so it never overwrites the notes you wrote by hand in ## 筆記區. This is a safety net. To regenerate from scratch, delete the old file first.

Mistake 4 — Passing Chinese: this skill only handles English. Give it a Chinese word and it will ask what you meant.

Next steps:

  • Want to auto-generate new words from your existing morphemes for practice? Try /new-word.
  • Want to verify that every two-way link in the dictionary still matches? Run /lint.
  • Want to add a freshly ingested word to your own review deck? Use /flashcard.
  • Want to change the output language (Chinese → English → Japanese…) or tweak other details? See Customize & tune.

Back to home.