Oculix
OculiX — Visual Automation IDE. If you can see it, you can automate it.
**The open-source reference for visual automation.** *Twenty years of lineage. MIT-licensed. Trusted in production by 91+ organizations worldwide.* The project is written primarily in Java, distributed under the MIT License license, first published in 2026. Key topics include: computer-vision, java, jna, nlp, ocr-recognition.
OculiX
The open-source reference for visual automation.
Twenty years of lineage. MIT-licensed. Trusted in production by 91+ organizations worldwide.
Get started · Documentation · Enterprise
</div>Trusted in production
| 91 | 14k+ | 3.4k | MIT |
|---|---|---|---|
| enterprise organizations | clones every two weeks | Maven downloads / quarter | open source, forever |
What OculiX is
OculiX drives any graphical interface by what it looks like — not by accessibility hooks, DOM selectors, or fragile XPaths. You point at what you see on screen, OculiX does the rest.
It works where selector-based tools cannot: native desktop applications, virtualized desktops (Citrix, RDP, VNC), Canvas/WebGL-rendered UIs, multi-application workflows, and any environment where what reaches the user is, ultimately, pixels.
You don't need to be a developer to use it. If you can take a screenshot, you can write your first OculiX script. If you are a developer, you get a full JVM scripting environment on top.
A first script
python# Open the app, export today's report, save it — every morning at 8 AM. from oculix import click, type, wait click("file_menu.png") click("export_to_csv.png") wait("save_dialog.png", timeout=10) type("filename_field.png", "report_today.csv") click("save_button.png")
Full installation, first-script tutorial, and tour of the IDE at oculix.org/getting-started.
What it brings
<table> <tr> <td width="50%">Pixel-perfect matching
OpenCV-based template matching with DPI awareness, similarity tuning, sub-pixel reliability. Five matching strategies cascaded before FindFailed.
OCR out of the box
Tesseract embedded via Legerix, no manual install. PaddleOCR available as opt-in HTTP server for multilingual and CJK workloads.
Cross-platform
Windows, macOS, Linux. One script runs everywhere — no platform-specific selectors. Native libraries pre-built and bundled.
Jython scripting
Python 2.7 syntax with full JVM interop. Drop-in compatible with legacy SikuliX scripts. JRuby, Robot Framework, and PowerShell runners also supported.
Full VNC stack
Connect to any remote machine without a local display. VNCScreen, VNCRobot, complete X keysym mapping, thread-safe parallel sessions.
Native SSH tunneling
SSHTunnel utility with embedded JSch. Open tunnels from Java alone — no shell wrapper, no WSL.
Android via ADB
Full ADB integration over WiFi or USB. No Appium, no XPath, no accessibility API. Validated on Android 12+.
Auditable execution
Optional MCP module: Ed25519-signed, SHA-256-chained JSONL audit journal. Every action independently verifiable.
How it compares
OculiX sits in the process automation space alongside RPA and visual automation tools.
| Capability | OculiX | UiPath | Eggplant |
|---|---|---|---|
| License & cost | MIT — free forever | Commercial — per bot / per user | Commercial — enterprise license |
| Self-hosted, fully local | 100% local, no cloud needed | Hybrid (Studio local, Orchestrator cloud) | On-prem available |
| Cross-platform | Windows · macOS · Linux | Studio is Windows-only | Windows · macOS · Linux |
| Approach | Pure visual — sees the screen | UI selectors (UIA/HTML) + visual fallback | Visual + IDE |
| Embedded OCR | Tesseract bundled | Multiple providers via marketplace | Built-in |
| Scripting language | Jython (Python 2.7 + JVM) | VB.NET · C# · Python (limited) | SenseTalk |
| Best fit | Anyone — non-tech users to enterprise teams | Enterprise RPA at scale | Visual test for regulated industries |
Comparison based on each vendor's public documentation as of 2026. OculiX is independent and unaffiliated with UiPath (UiPath, Inc.) or Eggplant (Keysight Technologies).
Quick install
Prerequisite: Java 11 or later — Eclipse Temurin or Azul Zulu.
Maven:
xml<dependency> <groupId>io.github.oculix-org</groupId> <artifactId>oculixapi</artifactId> <version>3.0.3</version> </dependency>
From source:
bashgit clone https://github.com/oculix-org/Oculix.git cd Oculix mvn clean install -DskipTests
Detailed setup at oculix.org/getting-started/installation.
Adopted by teams at
IBM · Qualcomm · Tencent · Alibaba · Kakao · Deutsche Bank · PepsiCo · Dassault Systèmes · Kaspersky · Synopsys · Neo4j · Zscaler · Jack Henry · RSA Security · Bill.com · University of Alberta
Organizations with verified OculiX downloads on Maven Central. Names listed for informational purposes only — no endorsement, partnership, or affiliation is implied. Full list on the adoption showcase.
Where to next
| Need | Where |
|---|---|
| Install OculiX | Get started |
| Your first script | Hello-world tutorial |
| OCR & text recognition | OCR guide |
| Bug reports & questions | GitHub Issues · Discussions |
| Security vulnerability | Private advisory — see SECURITY.md |
| Enterprise support | Work with us |
Contributing
Pull requests are welcome. See CONTRIBUTING.md for the workflow, code style, and review expectations. Substantive changes are discussed in advance via an issue when possible.
Visible at Contributors.
License & lineage
OculiX is released under the MIT License. It is the active continuation of the Sikuli (Tom Yeh, MIT UIST 2009) and SikuliX (Raimund Hocke, 2010–2025) lineage. The org.sikuli.* package namespace and historical prose are preserved as a matter of respect to the foundations.
🦎
Contributors
Showing top 12 contributors by commit count.
