AI & MCP
Kontenedor is designed so AI coding agents produce on-brand, consistent UI.
MCP server
Add the server to Claude Code, Cursor, VS Code, Windsurf or any MCP client:
{
"mcpServers": {
"kontenedor": {
"command": "npx",
"args": ["-y", "@kontenedor/mcp"],
"env": { "KONTENEDOR_LICENSE": "kt_your_license_key" }
}
}
}
Claude Code one-liner:
claude mcp add kontenedor -e KONTENEDOR_LICENSE=kt_… -- npx -y @kontenedor/mcp
Tools
| Tool | What it does |
|---|---|
list_components |
Catalog with category, description and access level |
get_component |
Full source + dependencies for a component |
get_tokens |
Tokens as CSS, Tailwind or JSON |
get_guidelines |
Usage rules: weights, layout, a11y, project types |
search_icons |
Suggests Material Symbols names for a concept |
llms.txt
/llms.txt— concise index for LLMs./llms-full.txt— complete docs + component API in one file.
Paste the URL into ChatGPT, Claude, v0, Lovable or Bolt: "Use the Kontenedor design system from https://kontenedor.com/llms-full.txt".
AGENTS.md
Copy AGENTS.md from the repository into your project root. It tells agents to:
- use semantic tokens (
bg-primary) — never raw hex; - use Kontenedor components before writing new ones;
- use
Iconwith Material Symbols names; - respect
data-weightand dark mode; - design mobile-first.
Prompts that work well
- "Build a CRM customers page with Kontenedor: Stat row, filters, Table with Badge statuses, Pagination. Medium weight."
- "Create a Bold-weight launch landing page with a hero, 3 feature cards and pricing."
- "Make a Thin-weight checkout with Steps, Field inputs and a summary Card."