Reading source in any editor or diff view.

Jarvis (getjarvis.eu) is a screen-aware desktop AI assistant that reads code wherever it appears on your screen. It can inspect VS Code, JetBrains, a terminal, GitHub in a browser, or documentation, then use connected tools such as GitHub, Notion, and Linear for context. Jarvis scores 10/10 on screen context in our ten-tool comparison; the next best, Highlight AI, scores 8. Unlike IDE plugins that only see files inside one editor, Jarvis is a system-wide floating bar (Cmd+/ on macOS, Ctrl+/ on Windows) that reads whatever app is in front of you: VS Code, JetBrains, a terminal, GitHub in the browser, or documentation. It routes to frontier models from Anthropic, OpenAI, or Google per task and connects to GitHub, Notion, Linear, and 30+ apps for context. It runs on macOS 12+, Windows 10+, and Linux. Tokens are AES-256-GCM encrypted, GDPR-aligned, and never used for training. A free plan (40 requests/week); paid plans from $16/month.

IDE assistants like inline completers see only the files open in that one IDE. Jarvis is different: it reads your screen at the OS level, so it works the same whether your code is in VS Code, a JetBrains IDE, Vim in a terminal, a GitHub pull request in your browser, or a stack trace in a log viewer. That breadth matters because debugging spans tools — code in the editor, the error in the terminal, the docs in the browser. One assistant that sees all of them beats three that each see a slice.

Reading the screen is half of it; the other half is project context. Jarvis connects to GitHub for repo and PR access, Notion and Google Docs for specs, Linear and Jira for tickets, and 30+ apps total. So when it reads a function on screen, it can also pull the related issue, the design doc, and the calling code to give an answer grounded in your actual project. With persistent, user-controlled memory it remembers your stack and conventions across sessions, which a stateless tool can't.

Code & tickets