what is stet fr?

Stet is an open source document editor that treats AI the way a human copy editor treats a manuscript. Instead of silently rewriting your prose in a chat window, the AI places visible marks on your text: highlights, circles, wavy underlines, and margin notes. You see each suggestion and decide whether to accept it, so the document only changes when you agree. The project is built with TypeScript and Next.js and runs as a local web app on your machine. The editor itself supports the kinds of elements you would expect from a rich text tool: headings, lists, quotes, code blocks, links, images, and page breaks. You also get human markup tools for highlighting, colored underlines, and circling passages, plus a notes panel for comments anchored to specific text. Documents can be viewed in continuous or paginated layouts (A4 or Letter), and there are several themes including light, dark, midnight, reading, and forest. Export options cover Markdown, Word, HTML, plain text, and PDF, and you can import from Markdown, text, HTML, and Word files. A central design choice is that your documents are plain files stored on your disk, in a folder called Stet inside your home directory by default. Each document is saved as a JSON file alongside a readable Markdown sibling, and folders become real subdirectories. There is no account, no server holding your data, and no telemetry. The app itself acts as the local server that reads and writes that folder, and if you deploy it to a host with no persistent disk, it falls back to storing documents in the browser instead. To use the AI review features, you paste your own API key from a provider such as Anthropic, OpenAI, or Gemini. The key is stored only in your browser's local storage and is sent directly to the provider you chose when you run a review. It never goes to a Stet server, because there is no Stet server. The only text that leaves your machine is what you send to your AI provider, under that provider's own privacy policy. The project is licensed under the GNU AGPL 3.0, which means anyone can use, study, self host, and modify it. If you run a modified version as a network service, you must share the source under the same terms. The maintainer reviews every change and asks that contributors open an issue to discuss ideas before sending a pull request.