About harukibox
harukibox is an MCP server for managing a proxy-buying business through its official Agent API. AI agents can work with product, order, buyer, and shipment data using read and write tools, with support for both newer and legacy MCP protocol paths.
Key features
- 01List, retrieve, and create products
- 02View orders, buyers, and shipments
- 03Search business data globally
- 04Supply tokens through environment variables
- 05Support newer and legacy MCP paths
Use cases
- 01Maintain a proxy-buying product catalog
- 02Check order and buyer information
- 03Review shipments in progress
- 04Search proxy-buying business records
MCP support
The MCP server gives AI clients harukibox tools for reading and creating products, viewing orders and buyers, listing shipments, and searching across business data.
Configuration
Remote servers need no local install. Pick your client, then copy the config or install in one click.
https://harukibox.com/api/agent/mcpclaude mcp add --transport http com-harukibox-harukibox https://harukibox.com/api/agent/mcp{"mcpServers":{"com-harukibox-harukibox":{"type":"http","url":"https://harukibox.com/api/agent/mcp"}}}Open in Cursor{"servers":{"com-harukibox-harukibox":{"type":"http","url":"https://harukibox.com/api/agent/mcp"}}}Open in VS Code{"mcpServers":{"com-harukibox-harukibox":{"type":"http","url":"https://harukibox.com/api/agent/mcp"}}}{"mcpServers":{"com-harukibox-harukibox":{"serverUrl":"https://harukibox.com/api/agent/mcp"}}}Run commands in a terminal; paste JSON into the client’s MCP config and restart. Servers that need sign-in will prompt on first connect.
Tools
Details
From the project README@harukibox/mcp
Model Context Protocol server for harukibox — let your AI agent manage your 代購 inventory, orders, buyers and shipments through the official Agent API.
協定版本
實作 MCP 2026-07-28(stateless、server/discover、resultType、cacheable
tools/list),同時維持 dual-era:舊版的 initialize handshake 照樣回應。
也就是說現在的 Claude Desktop / Claude Code(仍在 2025-11-25 以前)不用改任何設定 就能繼續用;等 client 升級到 2026-07-28 後會自動走新路徑,一樣不用改設定。
@modelcontextprotocol/sdk最新的 1.30.0 仍停在2025-11-25,沒有server/discover/resultType/CacheableResult,所以本 server 直接輸出 wire format 而不透過 SDK(也因此本套件零 runtime 相依)。SDK 支援新版後會換回去。
安裝
npm install -g @harukibox/mcp
或在專案內:
npm install @harukibox/mcp
取得 API token
透過 @harukibox/cli 跑 OAuth 取得(90 天有效):
npm install -g @harukibox/cli
harukibox login # PKCE,預設打開瀏覽器
# 或無 GUI 環境:
harukibox login --device # device flow,顯示 user_code 讓你在瀏覽器輸入
跑完 token 會寫進 ~/.config/harukibox/config.json(mode 0600)。把該 token 設成 HARUKIBOX_TOKEN 即可。
Claude Desktop 設定
編輯 ~/Library/Application Support/Claude/claude_desktop_config.json:
{
"mcpServers": {
"harukibox": {
"command": "harukibox-mcp",
"env": {
"HARUKIBOX_TOKEN": "hrk_live_your_token_here"
}
}
}
}
Refresh token (進階, 預設關閉)
如果你跑 long-lived MCP process(不常重啟),可開 refresh:
"env": {
"HARUKIBOX_TOKEN": "hrk_live_...",
"HARUKIBOX_ENABLE_REFRESH": "1",
"HARUKIBOX_REFRESH_TOKEN": "hrk_refresh_..."
}
⚠️ 限制:refresh rotation 後新 refresh 只在 process memory;process 重啟時 env 仍是舊 refresh,server 會視為 reuse 並撤該 user 所有 token。重啟前請手動執行 harukibox login 取新 token。建議 production 不開此選項,依賴 90 天到期後 re-OAuth。
Claude Code 設定
claude mcp add harukibox --command harukibox-mcp --env HARUKIBOX_TOKEN=hrk_live_your_token_here
可用工具
| Tool | 必要 scope | 說明 |
|---|---|---|
haruki_me | me | 取得當前 user/organization/token info |
haruki_list_products | products:read | 列出商品 |
haruki_get_product | products:read | 取得單一商品 |
haruki_create_product | products:write | 建立商品 |
haruki_list_registrations | registrations:read | 列出訂單 |
haruki_list_buyers | buyers:read | 列出買家 |
haruki_list_shippings | shippings:read | 列出出貨 |
haruki_search | search | 全域搜尋 |
安全
- Token 一律以環境變數提供,不會寫入磁碟
- 所有請求走 HTTPS 並帶 sha256 hash 驗證
- 過期 / 撤銷 / IP allowlist 不符時請求會立刻被拒
- 每次呼叫會被寫入伺服器端 audit log(沒有 web UI 查詢;如需匯出請與服務方聯繫)
自架 / 私有部署
HARUKIBOX_BASE_URL=https://your-instance.example.com/api/agent \
HARUKIBOX_TOKEN=hrk_live_... \
harukibox-mcp
License
MIT
Comments
New to MCP?
From what the protocol is to installing your first server, these answers cover most first questions.
An open protocol that defines how AI applications connect to external tools and data in a uniform way. Clients talk to the model; MCP servers provide tools, resources and prompt templates. They exchange standard messages, so one server can be reused by many clients.