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
Find more issues: in the issues popup, AI ▾ then Find more issues asks the AI provider
for problems LanguageTool missed in the whole text box. They show with a purple line and as
"AI suggestion" rows with the usual fix buttons and Ignore. Fixes Proofline applies keep the
other findings in place; editing the text yourself clears them.
More AI providers. Settings › AI now offers OpenAI, Gemini, Grok, Mistral, DeepSeek, Groq,
OpenRouter, Ollama (on this computer or another address you enter) and any OpenAI-compatible
server (LM Studio, llama.cpp, Jan, Together AI and similar), alongside Claude. One provider is
active at a time, and each keeps its own key, model and address. Providers other than Claude
and Ollama are marked "(untested)" until someone confirms them with a real account. The monthly spending
limit covers the cloud providers; Ollama and Other show tokens and requests instead. Copilot is
not supported: Microsoft and GitHub offer no public API for other apps. Existing settings keep
Claude.
Fixed
Text from the AI in the issues popup and the card is always shown as plain text.
One slow or failed reply from the LanguageTool server no longer shows the server error. Proofline
tries the check up to three times, one second apart, and shows the error only if all three fail.
Test in Settings › AI says "Ollama answered." (or "Other server answered.") when no key was sent,
instead of "The key works."