Documentation
Rosette documentation
Early development
Rosette is in early development. The API is unstable and will have breaking changes. Not suitable for production use.
What is Rosette?
Rosette is an AI-native, scriptable GDSII layout editor. It pairs a high-performance Rust core with an ergonomic Python API, so you can define your layouts in Python while the heavy lifting — geometry operations, routing, design-rule checks — runs at native speed. A web-based viewer with WebGPU rendering lets you inspect designs interactively, and a desktop app wraps it all into a single tool.
Rosette is built from the ground up to support agentic design workflows and ML-driven design models. Its Python-first API and structured layout representation make it straightforward to integrate with AI agents, optimization loops, and generative models — whether you are exploring a design space automatically or letting an agent handle routine layout tasks.
Whether you are drawing a simple test structure, assembling a complex multi-layer chip, or letting an AI agent iterate on your design, Rosette gives you programmatic control over every polygon while keeping the feedback loop fast.
Using these docs
The documentation is organized into three sections:
- Getting Started — Install Rosette, create your first layout, and learn the core concepts (cells, layers, ports, routing). Start here if you are new.
- Guides — Task-oriented walkthroughs that cover specific workflows: routing, design-rule checking, GDS-II import/export, and more.
- API Reference — Complete reference for every class and function in the Python API.
If you already know what you are looking for, use the search bar at the top of the page.
Pre-requisites
Rosette is designed to be approachable, but some background knowledge will help you get the most out of it:
- Basic Python experience — you should be comfortable writing and running Python scripts. No advanced knowledge is needed.
- Familiarity with layout design — an understanding of concepts like layers, cells, and design rules will help, but is not strictly required. The Getting Started section introduces the key ideas.
- Command-line basics — you will use the terminal to install Rosette and run your design scripts.
- AI agents and ML models (optional) — Rosette is designed to work with agentic workflows and ML-driven design tools. If you plan to use these features, familiarity with AI/ML frameworks and agent orchestration will be useful.
Join the community
Rosette is open source. If you run into a bug, have a feature request, or want to contribute, head over to the GitHub repository. Open an issue, start a discussion, or submit a pull request — all contributions are welcome.