Oculix

OculiX is an open source visual automation tool that drives any graphical interface by what it looks like, rather than through accessibility hooks, DOM selectors or XPaths. A user points at an image of a button or window, and the script clicks, types and waits on it. It is a fork of the long running SikuliX project and is MIT licensed. It includes a scripting IDE with Python 2.7 syntax through Jython and full JVM interop, and it stays compatible with older SikuliX scripts. Matching works by sample image with tunable similarity, and a vision pipeline offers template, feature and contour primitives. Tesseract OCR is embedded, with PaddleOCR available as an optional HTTP server for Chinese, Japanese, Korean and complex layouts. The same API can drive remote machines over VNC, open SSH tunnels from Java, and control Android phones over ADB. It runs on Windows, macOS and Linux. The AI connection is an MCP server, oculixmcp, which exposes the engine's click, type, find, OCR and screenshot actions to Model Context Protocol clients including Claude Desktop, Claude Code, Cursor, Windsurf, Gemini CLI, VS Code and Goose. Each action is signed and written to a hash chained audit journal, so the server is aimed at controlled, auditable environments, with an optional confidential mode and token protection. Typical uses are repetitive workflows such as exporting reports and filling the same form many times, recording demos and tutorials, and triggering custom actions for accessibility needs. The software is free, and the site lists no paid plans. A separate enterprise support page covers getting help.