Description
Caret provides a unique design layer that lives directly within your code repository, offering a free and open-source solution for developers. Unlike traditional design tools that separate design files from the codebase, Caret keeps your design as real React components in a `.caret/` directory. This means your design is versioned by Git, reviewable in pull requests, and synced into your application only when you choose.
Read More
The tool addresses common pain points in the design-to-code workflow, such as design mockups drifting out of sync with the actual build, repetitive fixes being overwritten by generated changes, and the difficulty of describing visual changes with words. With Caret, your design is the code itself, ensuring consistency and persistence.
Key capabilities include direct editing on a live canvas, where users can right-click and modify text, colors, or images, with changes immediately reflected in the `.tsx` files. It intelligently uses your existing design tokens, writing token references instead of raw hex codes, which simplifies brand color updates. For more complex changes, users can paint over a section of the page and describe the desired outcome, allowing an AI agent to target exact elements.
Caret also supports generating multiple design versions side-by-side for exploration and creating assets like vector logos, photographs, and animated backgrounds that adhere to your token system for consistent styling. It provides a comprehensive view of your product, rendering the active page as live, interactive React, while other pages remain fast thumbnails. Viewport presets allow for checking responsiveness across different screen sizes.
The application runs locally on your machine and requires no account, ensuring your code and designs remain private. It integrates with AI coding agents like Claude Code and Cursor by keeping design tokens current in their respective rules files (`CLAUDE.md`, `.cursor/rules`, `AGENTS.md`) and enabling agents to perform the design-to-app sync via a local MCP server. Caret supports various application frameworks like Next, Remix, Vue, Svelte, and Rails, as the design layer is always React, and sync translates it to your chosen stack.
Caret is designed to complement existing AI tools and workflows, rather than replace them. It works alongside tools like Anthropic's frontend-design skill and Figma MCP, providing the crucial persistence layer that keeps design decisions grounded in your repository. It's available for macOS, Windows, and Linux, requiring Node.js for the live canvas.
- CategoryAI Coding Assistants
- Pricing ModelFree
- DeploymentLocal
- PlatformsmacOS, Windows, Linux
Caret's Core Features
Design layer in your repo as real React
Live canvas for visual editing
Directly edit text, colors, and images on page
Writes changes into source files (.tsx)
Uses design tokens for consistent styling
AI agent integration for complex changes
Generates multiple design versions side-by-side
Creates assets (vector logos, photos, animated backgrounds)
Full product view with live interactive pages and thumbnails
Viewport presets for responsive design checks
Runs locally on your machine
No account required
Integrates with Claude Code via MCP
Integrates with Cursor via MCP
Keeps design tokens updated in agent rule files
How to use Caret?
Install Caret: Download and install the desktop application for macOS, Windows, or Linux.
Open Project: Open an existing project folder in Caret to create the `.caret/` design layer.
Edit Visually: Right-click on elements like text, colors, or images on the live canvas to make direct changes.
Use AI Agent: For complex adjustments, paint over a page region and describe the desired change to your connected AI agent.
Manage Tokens: Caret automatically manages and updates your design tokens in agent-specific rule files.
Sync Design: When ready, initiate a sync to apply design changes from the `.caret/` folder to your main application code using your AI agent.
Caret's Use Cases
- Maintaining Design-Code Sync
- Visual UI Editing for Developers
- Version-Controlled Design
- AI-Assisted Design Refinement
- Consistent Design System Application
- Rapid Design Prototyping
- Automated Asset Creation
- Cross-Framework Design Sync







