Skip to main content
@loftlyy/mcp exposes Loftlyy brand data through the Model Context Protocol over stdio.

​
What the server provides

  • Read-only tools for brand lookup, search, filtering, similarity, palettes, and facets
  • A brand://{slug} resource for full brand profiles
  • Prompt templates for branding guides, brand comparisons, and palette suggestions

​
Run the published server

npx -y @loftlyy/mcp

​
Client setup

​
Claude Desktop

Add this server to claude_desktop_config.json: 
{
  "mcpServers": {
    "loftlyy": {
      "command": "npx",
      "args": ["-y", "@loftlyy/mcp"]
    }
  }
}

​
Claude Code

claude mcp add loftlyy -- npx -y @loftlyy/mcp

​
Cursor or another stdio client

Configure a stdio MCP server that runs: 
npx -y @loftlyy/mcp

​
Environment variables

VariableDescriptionDefault
LOFTLYY_SOURCEForce local or remote mode.Auto-detected
LOFTLYY_BASE_URLOverride the remote API base URL.https://loftlyy.com
LOFTLYY_ROOT_DIRRoot directory used for local brand file discovery.process.cwd()
LOFTLYY_STRICTSet to true to fail on invalid brand data.false

​
Local repo mode

If the server runs inside the Loftlyy repository, it auto-detects local mode when data/brands/ exists. If the server runs outside the repo, point it back to your checkout: 
{
  "mcpServers": {
    "loftlyy": {
      "command": "node",
      "args": ["/path/to/loftlyy/mcp/dist/index.js"],
      "env": {
        "LOFTLYY_SOURCE": "local",
        "LOFTLYY_ROOT_DIR": "/path/to/loftlyy"
      }
    }
  }
}

​
First prompts to try

After the server is connected, try prompts like these in your client:
  • What colors does Stripe use?
  • Find brands similar to Airbnb.
  • Show me technology brands with blue palettes.
  • Generate a branding guide for a fintech startup inspired by Stripe and Linear.

​
Development

Build the server locally: 
pnpm mcp:build
Test it with the MCP Inspector: 
LOFTLYY_SOURCE=local \
LOFTLYY_ROOT_DIR=/path/to/loftlyy \
npx @modelcontextprotocol/inspector node /path/to/loftlyy/mcp/dist/index.js