What it does
It runs only when explicitly invoked and does not rewrite or block text. Its style-impression label does not establish who wrote a passage, and mock tests do not demonstrate real-provider accuracy.
Integrations
A Hermes plugin that reads Japanese prose in context and offers eight separate labels to help a person decide what needs attention.
Only you can see your notes.
Screenshot unavailable. Open the experiment ↗
It runs only when explicitly invoked and does not rewrite or block text. Its style-impression label does not establish who wrote a passage, and mock tests do not demonstrate real-provider accuracy.
A developer can borrow this method for a Japanese proofreading tool. The code splits a document into separate sentences. It gathers the surrounding text for context. Then, it sends small groups of sentences to an outside AI service like TypeSafe. The service returns labels highlighting unclear or redundant writing.
To connect your app, you need an access key from TypeSafe or OpenRouter. The app sends your Japanese text to this outside provider to be read. You should not share private documents. The tool offers subjective advice for a person to review. It never rewrites the text automatically.