Fragiola

Getting Started

AI agents

Give your AI agent Fragiola's docs: the Fragiola MCP server answers what exists and how to install it, and the agent skills teach how to build with it.

Fragiola publishes two tools for AI agents (Claude Code, Cursor, VS Code, Codex, OpenCode, …). They work in any project and are independent of each other:

toolwhat it gives the agentinstall
MCP server (the fragiola CLI)everything fragiola.com documents: which components exist, their docs, the libraries' APIs and how to install eachnpx fragiola@latest mcp init --client <client>
Agent skills (fragiola/skills)how to build with Fragiola: palettes, style families, composition, the headless libraries, the MCP servernpx skills add fragiola/skills

The MCP server

Fragiola has two kinds of components, and the server always says which one an item is:

  • Fragiola UI: components as source code, installed with npx shadcn@latest add @fragiola/<name>. The code is copied into your project, which owns it.
  • Libraries: headless npm packages, a core @fragiola/<x> and adapters such as @fragiola/<x>-react. You install the adapter for your framework and import from it. The special components are the styled layer over these.

Every answer comes from fragiola.com and is fetched when a tool needs it, so new components and pages show up without upgrading anything.

Set it up

Run this in your project:

npx fragiola@latest mcp init --client <client>
client--clientwhat it does
Claude Codeclaudewrites .mcp.json. Restart Claude Code and approve the server (/mcp shows it)
Cursorcursorwrites .cursor/mcp.json. Enable the server in Cursor Settings → MCP
VS Codevscodewrites .vscode/mcp.json. Click Start next to the server, then use Copilot Chat in agent mode
OpenCodeopencodewrites opencode.json. Restart OpenCode
Codexcodexprints the block for ~/.codex/config.toml. Codex keeps its servers outside the project, so nothing is written

Without --client, it asks which client to configure. It merges into an existing config file and keeps your other servers, and running it again changes nothing. fragiola is not added to your dependencies: every client runs npx fragiola@latest mcp, so it always uses the latest server.

Configure it by hand

Claude Code (.mcp.json) and Cursor (.cursor/mcp.json):

{
    "mcpServers": {
        "fragiola": { "command": "npx", "args": ["fragiola@latest", "mcp"] }
    }
}

VS Code (.vscode/mcp.json):

{
    "servers": {
        "fragiola": { "command": "npx", "args": ["fragiola@latest", "mcp"] }
    }
}

OpenCode (opencode.json):

{
    "$schema": "https://opencode.ai/config.json",
    "mcp": {
        "fragiola": {
            "type": "local",
            "command": ["npx", "fragiola@latest", "mcp"],
            "enabled": true
        }
    }
}

Codex (~/.codex/config.toml):

[mcp_servers.fragiola]
command = "npx"
args = ["fragiola@latest", "mcp"]

The server needs Node.js 24 or later on the client's PATH.

Ask it

  • "Which Fragiola components exist for forms?"
  • "Add a Fragiola select to this form, styled with the palette roles."
  • "Which Fragiola library gives me dockable panels, and how do I install it?"

The agent skills

Agent skills are instructions an agent loads when a task needs them. The fragiola skill teaches it to build with Fragiola UI, the headless libraries and the MCP server. Install it from your project with the skills CLI:

npx skills add fragiola/skills

Add -g to install it for every project, or -a <agent> to pick the agent. You can also copy skills/<name>/ from the repository into your agent's skills folder (.claude/skills/, .agents/skills/, …).