v0 Unveils Design Systems 2.0 with Multi-Source Import and Usage Intelligence.

v0 announced on LinkedIn that Design Systems 2.0 is now available, expanding how teams import and use their component libraries inside the AI code generation platform. The update adds support for importing from GitHub repositories, public and private npm packages, Storybook documentation, Figma frames, screenshots, ZIP files, and live applications.

The new version analyzes not just component structure but also how design systems are used in production. v0 now extracts components and props, design tokens and themes, providers and global styles, plus usage patterns from actual codebases. Once imported, the system generates an interactive playground where teams can preview components with their real tokens, iterate through chat, and save the result when ready.

This follows v0's recent expansion into visual editing capabilities and reusable agent skills, part of a broader push to make the platform handle more complex, real-world development workflows.

Teams can pull design systems from sources they already maintain. GitHub repository imports scan codebases for component definitions and usage patterns. npm package imports work with both public registries and private packages, routing authentication through shared environment variables rather than asking for secrets in chat. Storybook and documentation imports parse component APIs and examples.

The visual import options handle Figma frames, screenshots of existing UIs, ZIP archives of code, and URLs of deployed applications. v0 analyzes these inputs to understand component structure, styling conventions, and how pieces fit together. The system learns not just what components exist but how developers actually compose them, which props are commonly used together, and what patterns appear across multiple implementations.

After import, v0 creates a playground specific to that design system. The playground loads with the team's components and tokens already configured. Developers can test combinations, refine through conversational prompts, and verify the output matches their standards before committing the design system to their workspace.

Once a design system is imported and saved, it becomes available across the team's workspace. Developers can select it from the prompt bar when starting a new generation, begin from the Design Systems page, or have it load automatically based on project context. The same design system works across all chats, maintaining consistency as different team members build features.

The changelog notes that design system selection now persists per scope, so switching between projects remembers which system each uses. This removes the need to reselect design systems when moving between different codebases or client projects.

Private npm package authentication runs through environment variables configured at the team level, avoiding the security risk of pasting tokens into chat messages. This change shipped as part of broader improvements to how v0 handles secrets across the platform.

The same release includes Platform API v2 in beta, which can now start chats directly from GitHub repositories. The API documentation adds a migration guide for teams moving from v1 and a version switcher for developers working across both.

v0's MCP server adds tools to create, find, and read chats; fetch previews; and send messages programmatically. File attachment support expands to include Excel and Word documents. Pasted images and files now appear as inline chips in the prompt composer with thumbnails, and remain visible across page reloads.

Custom skills can now set disable-model-invocation to prevent automatic inclusion in the agent's tool catalog while staying available when explicitly attached or loaded. This gives teams finer control over which capabilities appear in different contexts without removing them from the workspace entirely.