SikuliX uses computer vision (OpenCV) to identify and interact with anything visible on a screen — Windows, macOS, Linux. It locates GUI elements through image recognition, then drives them with simulated mouse and keyboard input. No access to source code, DOM, or accessibility APIs required.

The original motto, still valid : "If you can see it, you can automate it."

For the modern continuation with OpenCV 4.10 (via Apertix), Java 17+, working VNC stack, Android 12+ ADB, PaddleOCR option, and 22 native-reviewed locales → oculix-org/Oculix.

A 23-year lineage of MIT-licensed open work :

• 2003 · MIT CSAIL — Rob Miller's UI Design Group launches the visual automation research project that would become Sikuli.
• 2009 · UIST paper — @doubleshow (Tom Yeh) and @vgod (Tsung-Hsiang Chang) formally introduce Sikuli at the ACM Symposium on User Interface Software and Technology. The codebase eventually lands under the sikuli GitHub organization.
• 2010–2023 · SikuliX1 — @RaiMan (Raimund Hocke) takes over as sole maintainer, evolving the project through Java 8 → 11, multi-platform fixes, OCR integration, and 13 years of patient stewardship.
• March 2026 · oculix-org — RaiMan archives the upstream and transmits stewardship; active development continues as OculiX.

Massive gratitude to all four — without them, this project simply doesn't exist. The MIT license they chose 23 years ago is still in place today, and that's not an accident.