You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Optional AI suggestions through Claude, Anthropic's AI model, with your own API key. Off by
default. Settings › AI holds the on/off switch, the key (kept in Windows Credential
Manager) with a Test button, the model, a monthly spending limit (default $1.00), and
today's and this month's totals.
AI Rephrase in the card under an underlined word: up to three rewordings of that sentence,
changed words in bold. Choosing one replaces only that sentence. Only the sentence is sent.
A shortcut to rephrase the sentence at the caret, for sentences with no underlined word. Set
it in Settings › Shortcuts (none by default).
An AI Rephrases badge in the README: the number of AI rewordings applied, counted like
Corrections Applied and turned off by the same setting.
Wavy red lines under flagged words, drawn over the text box in apps that report where each
word is on screen (tested in Edge and Notepad). They follow the text when it scrolls, disappear
while you type, and return after the next check. Clicks go through to the text box.
Settings › Appearance › Underline issues in the text turns them off.
A card under an underlined word while the caret is inside it: up to three suggestions (the
first highlighted), Ignore, Add to dictionary for spelling, and All issues… for the popup.
Hidden characters are flagged: a zero-width space, soft hyphen or direction mark inside a word,
text-direction controls anywhere, and unusual spaces (thin, em, narrow no-break, ideographic and
similar). Placeholders that web editors (Slate, Quill, CKEditor 5) insert themselves are not. The
popup shows each as its code, such as [U+200B], with a fix that removes it or puts a normal
space in its place. On by default; Settings › Checking › Flag hidden characters turns it
off. LanguageTool does not flag these itself.
Fixed
A password field is recognised as soon as it has focus, before Proofline looks at the
text box around it, so its text is never read.
A suggestion that removes text ((remove)) now deletes the selection with the Delete key
instead of pasting an empty clipboard.