Character and word counts are the obvious ask, but the profile is what you cannot get from a word processor: sentence length, part-of-speech mix and type-token ratio turn "this reads heavy" into numbers you can act on. It measures — it does not rewrite — so use it as feedback on drafts.
A morphological analysis server for Japanese text, built on kuromoji.js. It counts characters and words the way Japanese actually requires — words are not space-separated — and reports linguistic features you can use as writing feedback. It works on pasted text or on a file, and resolves loose file paths across several common directories.
- Character counts excluding spaces and line breaks, from a file or from pasted text — `count_chars`, `count_clipboard_chars`
- Word counts that use morphological analysis for Japanese and space splitting for English, selected with a `language` parameter — `count_words`, `count_clipboard_words`
- A full linguistic profile: average sentence length, morphemes per sentence, part-of-speech proportions, particle usage, the hiragana/katakana/kanji/alphanumeric mix, vocabulary diversity as a type-token ratio, loanword share, honorific frequency and punctuation per sentence — `analyze_text`, `analyze_file`
- Path handling that accepts Windows and WSL-style absolute paths, relative paths, or just a filename, searching the working, home, desktop and documents directories
No account and no key. Run it straight from GitHub with `npx -y github:Mistizz/mcp-JapaneseTextAnalyzer`, or install into Claude Desktop through Smithery; on Windows, wrapping the command in `cmd /c` is the documented fallback. The package is `mcp-japanesetextanalyzer` 1.0.0. The morphological dictionary loads at server startup, which is why the first run takes a moment.
One command — npx -y github:Mistizz/mcp-JapaneseTextAnalyzer