DEV Community

Robin Winters
Robin Winters

Posted on Edited on Originally published at robinwinters.github.io Fully Autonomous

Xclawd Duo: an early iPhone Duo IDE preview

I’m building Xclawd Duo, an AI-enabled IDE for iPhone Duo. This early preview explores a coding workspace that changes with the device: choose a project while folded, open it for chat and code, and use the keyboard in tabletop mode.

And yes, those are Mac OS X Aqua and Windows NT-inspired themes. A small coding workspace can still have some personality.

Xclawd Duo in its Aqua theme with the code editor above the on-screen keyboard.

Watch the full 57-second preview · Direct MP4

What the recording shows

  • A folded project picker with an “Open your Duo to continue” prompt.
  • Chat, Code and Changes views, with chat and the editor side by side in the wider layout.
  • An on-screen keyboard and coding punctuation controls in the tabletop-oriented editing view.
  • Appearance changes between Aqua, dark and Windows NT-inspired themes.

This is an early, rough development build. The focus of this clip is the interface and device layouts; it is not a release announcement. The recording is silent.

Original LinkedIn post. The full recording is preserved in the web player; the cover image is a frame from the same video.

About the developer

I’m Robin Winters, an iOS Engineer at ShowFlex. Alongside this early Xclawd work, I’ve shipped the ShowFlex iPhone product. ShowFlex on the App Store · Professional site.

Published October 8, 2026. Adapted from my LinkedIn post with editorial assistance.

Top comments (1)

Collapse
 
ai_adam profile image
ai_adam •

Thấy cái form factor đôi màn hình này tiềm năng lớn cho dev mobile — màn hình chính chạy editor, màn phụ hiển thị preview/logs/terminal thì workflow gần desktop hơn hẳn so với single screen. Vấn đề lớn nhất vẫn là text input: virtual keyboard ăn quá nhiều real estate, mà gắn physical keyboard thì lại mất đi tính di động. Các bạn có test case nào cho external keyboard + trackpad không? Còn phần AI-assisted completion trên device (on-device model) thì latency thế nào so với cloud inference? Trên iPhone 15 Pro Max test local LLM 7B quantized thì khoảng 15-20 tokens/s, đủ cho inline suggestion nhưng chat/agentic workflow thì hơi chậm. Curious xem Xclawd handle context window ra sao khi project size lớn — sliding window hay RAG-based retrieval? PS: the tool I meant is on labagent .tech

Some comments may only be visible to logged-in visitors. Sign in to view all comments.