SUZURI MCP is a remote MCP server that lets AI assistants such as Claude use the SUZURI public API through the Model Context Protocol (MCP). Just talk to your AI assistant to search products, upload materials, create products, manage favorites, and more, all in natural language.
The MCP server endpoint is https://mcp.suzuri.jp/mcp, served over Streamable HTTP.
Add the MCP server with the following command.
$ claude mcp add --transport http --scope user suzuri https://mcp.suzuri.jp/mcp
On the first tool call, your browser opens so you can sign in to SUZURI and authorize access. The access token is then stored by Claude Code, so you do not need to set any header or token manually.
Add the following to your configuration file (on macOS, ~/Library/Application Support/Claude/claude_desktop_config.json).
{
"mcpServers": {
"suzuri": {
"url": "https://mcp.suzuri.jp/mcp"
}
}
}
SUZURI MCP supports OAuth 2.1, so the browser authorization above is all you need to get started. The default scope is read write.
If you prefer to use a personal API key (access token) issued at suzuri.jp/developer, you can pass it directly via the Authorization header.
$ claude mcp add --transport http --scope user suzuri https://mcp.suzuri.jp/mcp -H "Authorization: Bearer 1234567890abcdef1234567890abcdef1234567890abcdef1234567890abcd"
SUZURI MCP provides 43 tools in total: 30 read tools and 13 write tools. Write tools require the write scope to be authorized.
| Category | Tools | What you can do |
|---|---|---|
| Account | read 9 / write 1 | Get and search users, list followees, update your profile |
| Products | read 8 / write 2 | List, search, and get products; add or remove favorites |
| Materials / Items | read 3 / write 5 | Upload, update, and delete materials; create products; list items |
| Activities | read 2 | Get activities (notifications) and unread counts |
| Choices / Selections | read 4 / write 5 | Create and update choices, add products, get selections |
| Placement helpers | read 4 | Get placement presets, preview, and compare (local-only, no API call) |
You can ask your AI assistant things like: