Smart Autocorrect
approvedby Zangeti
Phone-style autocorrect and next-word prediction directly in your vault. - This plugin has not been manually reviewed by Obsidian staff.
✍️ Smart Autocorrect for Obsidian
Phone-style autocorrect and next-word prediction, right inside your vault.
It runs completely offline, so your notes stay private and never leave your device.
⚡ As you type
- Real-time autocorrect. Typos are fixed automatically as you write.
- Next-word and phrase completion. A popup predicts what comes next. Press
Tabto accept. - Contextual suggestions. It reads the sentence you're writing and your other notes.
- 100% local. Everything runs on-device. No cloud, no telemetry, no account.
✨ Additional features
- Suggest alternatives. Right-click any word for more eloquent wording (good becomes substantial, help becomes facilitate, big becomes immense).
- Smart capitalisation. Sentence starts, names and places, and abbreviations like
e.g.andU.S.are handled for you. - Learns your writing. Words from your own notes come up first. Made a wrong correction? Just undo with
Ctrl/Cmd+Z: it restores your text, remembers the word, and stops correcting it. - Note links (experimental). Text that matches one of your notes gets underlined; hover to preview, click to link.
- Writing stats. See keystrokes saved, typing time saved, and your daily streak.
🚀 Getting started
- Install and enable the plugin.
- Accept the one-time model download (86 MB) when prompted.
- Start typing. There's nothing to configure; everything in settings is optional.
🔒 Privacy
100% local, with no telemetry and no cloud. The only network request is the one-time model download, which you can decline. Anything the plugin learns lives in personalization.json inside the plugin folder, so it travels with your vault and never touches your notes.
Supported languages: English only. If you'd like another language, please open an issue.
☕ Support
If this saves you time, you can buy me a coffee.
Credits
Started as a fork of Various Complements by Tadashi Aikawa, since rewritten around a local neural language model. MIT licensed.
For plugin developers
Search results and similarity scores are powered by semantic analysis of your plugin's README. If your plugin isn't appearing for searches you'd expect, try updating your README to clearly describe your plugin's purpose, features, and use cases.