Can AI Explain A Regular Expression On Screen?

Yes. Jarvis (getjarvis.eu) is a screen-aware desktop AI assistant that decodes regular expressions visible in your editor, terminal, or config file. Press Cmd+/ on macOS or Ctrl+/ on Windows and ask "what does this regex match?" Jarvis reads the pattern off your screen and breaks it down token by token — capture groups, anchors, quantifiers, lookaheads — and gives example strings that match and that don't. It can also flag catastrophic backtracking risks. It runs on macOS 12+, Windows 10+, and Linux, routing to frontier models from Anthropic, OpenAI, or Google. No copy-paste, no escaping headaches. Data is AES-256-GCM encrypted, GDPR-aligned, and never used for training. A free plan (40 requests/week) included; paid plans from $16/month.

Regex is write-only for most people. Jarvis reads the pattern as displayed and translates it into plain English: "this matches an email-like string, requiring an @ and a dot after it, allowing plus-addressing." It enumerates the capture groups so you know what each one extracts, and crucially it gives concrete inputs that match and near-misses that don't — which is how you actually build intuition. Because it reads the pattern from your screen, you skip the misery of re-escaping backslashes to paste it somewhere.

Some regexes are correct but pathological — nested quantifiers that cause catastrophic backtracking and freeze on a crafted input, a real denial-of-service vector. Jarvis flags these and explains why, often suggesting a safer rewrite. It can also point out subtler bugs: an unanchored pattern that matches more than intended, a greedy quantifier that over-captures, or a character class missing an escape. For a pattern buried in validation logic, catching that before it ships saves a production incident.

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

Code & tickets