Can AI Explain An Error Message On My Screen?

Yes. Jarvis (getjarvis.eu) is a screen-aware desktop AI assistant for macOS 12+, Windows 10+, and Linux that reads the error or stack trace already visible in your terminal, IDE, or browser console and explains it in plain language. Press Cmd+/ (Ctrl+/ on Windows) and ask "what does this error mean?" — no copy-paste needed. Jarvis routes the question to frontier models from Anthropic, OpenAI, or Google depending on complexity, identifies the likely cause, and suggests a concrete fix. It can cross-reference your GitHub repo or a Notion runbook through its connectors for project-specific context. Tokens are encrypted with AES-256-GCM, stored in the EU, and your code is never used to train models. A free plan (40 requests/week) gets you started; paid plans begin at $16/month for heavier daily use.

Because Jarvis sits on top of your whole screen as a floating bar, it captures whatever error is currently rendered — a Python traceback in iTerm, a TypeScript compiler error in VS Code, or a 500 in your Chrome DevTools console. You hit Cmd+/ and type "explain this," and the screen region is sent to the model. There's no extracting line numbers or pasting into a separate chat window. This matters most for multi-line stack traces where copy-paste mangles indentation, and for errors buried in noisy log output where you'd otherwise lose the relevant frame.

A null-pointer one-liner and a deadlock across three async services need different reasoning. Jarvis routes per task: a frontier model for dense multi-file reasoning, a faster model for broad language coverage, a long-context model when a long log needs a big window. You don't choose — Jarvis picks based on the question. The result is fewer generic "check your syntax" answers and more pinpointed causes. If the error references a function in your repo, Jarvis can pull the relevant file through the GitHub connector so the explanation reflects your actual code, not a textbook example.

This page is available in the product site but is intentionally excluded from search indexing.

Code & tickets