OpenPencil

OpenPencil is a design tool open-sourced by GitHub user ZSeven-W, and its README calls it "The world's first open-source AI-native vector design tool." On an infinite canvas you ask AI to create or change a design with prompts, and you can have several AI agents draw different pages at the same time, a feature the project calls Agent Teams. Designs are saved as JSON .op files and can be exported as React, HTML, Vue, SwiftUI and other code. Its most unusual feature is that several AI agents can split up the pages on one shared canvas. It runs in the browser and on macOS, Windows and Linux, with the interface in 15 languages including Simplified Chinese.

Who it's for:people designing interfaces and product prototypes, and front-end developers who want designs turned straight into code.

Before you start: the AI features need your own model service key, such as Claude, GPT, Gemini, Kimi, GLM or DeepSeek, and your prompts and canvas content go to the service you choose. If you use an aggregator service, your key and content pass through one more third party. A built-in MCP server can read and write design files and can be installed into an AI agent in one click. The code is free under the MIT licence; model calls may be billed by usage. The README includes sponsor promotion links, and we found no separate privacy policy.

Based on the project's public documentation. Not tested by AI Inspiration World.