user-majico-mcp
MCP server for Majico.xyz that enables coding agents to read (and limited write) brand guidelines, design tokens, studio canvas, and export manifests.
README
user-majico MCP server
MCP server for Majico.xyz: read (and limited write) tools so coding agents can pull brand guidelines, design tokens, studio canvas, and export manifests.
Requirements: Majico account with tokens. Cursor humans connect via OAuth (no API key in mcp.json). API keys are for automation/stdio only.
Install
npx -y user-majico-mcp@0.4.1
Environment variables (v0.2+)
Set once in MCP config — tools do not require projectId / apiKey on every call:
| Variable | Description |
|---|---|
MAJICO_API_URL |
API base (default http://localhost:3000; production https://api.majico.xyz) |
MAJICO_PROJECT_ID |
Project UUID |
MAJICO_API_KEY |
Project API key |
Per-call overrides in tool arguments still work.
Tools
| Tool | Description |
|---|---|
get_brand_profile |
Archetypes, niche intent |
get_design_tokens |
Palette and fonts |
get_logo_svg |
Selected logo SVG |
get_guidelines |
Markdown + LLM prompt |
get_design_md |
DESIGN.md markdown |
get_studio_canvas |
Studio canvas snapshot |
get_export_manifest |
Export file list |
download_export_zip |
Full BRAND.md / DESIGN.md / tokens ZIP (base64 in MCP response) |
submit_brief |
Submit product brief + enqueue niche research |
update_studio_html_frame |
Patch one htmlFrame (requires elementId, html) |
generate_asset |
Enqueue harness jobs (landing-page, investor-pack, investor-one-pager, investor-data-room, …) |
get_asset_status |
Poll harness job status |
Cursor agent example: generate_asset for investor pack
After brand chain, palette, GTM, team, and traction are ready:
{
"tool": "generate_asset",
"arguments": {
"skillId": "investor-pack",
"params": {
"team": {
"members": [
{ "name": "Alex Kim", "role": "CEO", "bio": "ex-Stripe PM" }
]
},
"traction": {
"metrics": [{ "label": "Waitlist", "value": "1,200" }]
},
"includeSlides": ["financials"]
}
}
}
Poll with get_asset_status({ jobId }). Preflight failures return HTTP 422 with code: "investor_pack_preflight_blocked" and a blocked reason (brand, gtm, team, traction, …).
Phase 2: one-pager and data room (after deck exists on canvas)
{
"tool": "generate_asset",
"arguments": {
"skillId": "investor-one-pager"
}
}
{
"tool": "generate_asset",
"arguments": {
"skillId": "investor-data-room"
}
}
Both require a harness-investor-pack element on the Studio canvas (generate investor-pack first). Without a deck, jobs return a stub with canvasUrl pointing to finish the deck.
Cursor .cursor/mcp.json (OAuth — preferred)
{
"mcpServers": {
"majico": {
"url": "https://api.majico.xyz/mcp"
}
}
}
Complete Connect in Cursor Settings → MCP. Use list_projects and pass projectId on tool calls to switch scope.
Cursor .cursor/mcp.json (stdio — automation / local)
{
"mcpServers": {
"majico": {
"command": "npx",
"args": ["-y", "user-majico-mcp@0.4.1"],
"env": {
"MAJICO_API_URL": "https://api.majico.xyz",
"MAJICO_PROJECT_ID": "<uuid>",
"MAJICO_API_KEY": "<key>"
}
}
}
}
HTTP MCP (hosted)
OAuth (Cursor humans): https://api.majico.xyz/mcp — Connect in Cursor; no headers in mcp.json.
API key (automation only):
Authorization: Bearer <project_api_key>
X-Majico-Project-Id: <uuid>
Development
cd packages/user-majico-mcp
npm install
npm run build
npm test
License
MIT
推荐服务器
Baidu Map
百度地图核心API现已全面兼容MCP协议,是国内首家兼容MCP协议的地图服务商。
Playwright MCP Server
一个模型上下文协议服务器,它使大型语言模型能够通过结构化的可访问性快照与网页进行交互,而无需视觉模型或屏幕截图。
Magic Component Platform (MCP)
一个由人工智能驱动的工具,可以从自然语言描述生成现代化的用户界面组件,并与流行的集成开发环境(IDE)集成,从而简化用户界面开发流程。
Audiense Insights MCP Server
通过模型上下文协议启用与 Audiense Insights 账户的交互,从而促进营销洞察和受众数据的提取和分析,包括人口统计信息、行为和影响者互动。
VeyraX
一个单一的 MCP 工具,连接你所有喜爱的工具:Gmail、日历以及其他 40 多个工具。
graphlit-mcp-server
模型上下文协议 (MCP) 服务器实现了 MCP 客户端与 Graphlit 服务之间的集成。 除了网络爬取之外,还可以将任何内容(从 Slack 到 Gmail 再到播客订阅源)导入到 Graphlit 项目中,然后从 MCP 客户端检索相关内容。
Kagi MCP Server
一个 MCP 服务器,集成了 Kagi 搜索功能和 Claude AI,使 Claude 能够在回答需要最新信息的问题时执行实时网络搜索。
e2b-mcp-server
使用 MCP 通过 e2b 运行代码。
Neon MCP Server
用于与 Neon 管理 API 和数据库交互的 MCP 服务器
Exa MCP Server
模型上下文协议(MCP)服务器允许像 Claude 这样的 AI 助手使用 Exa AI 搜索 API 进行网络搜索。这种设置允许 AI 模型以安全和受控的方式获取实时的网络信息。