mcp-dashboard
Interactive dashboard that works as an MCP server, allowing Claude Code to create and manage workspaces, widgets, tasks, and reminders in real-time via Socket.io.
README
MCP Dashboard
Interactive dashboard that works as an MCP server. Claude Code (or any MCP client) calls tools to create workspaces, widgets, and artifacts in real-time. The browser renders everything live via Socket.io.
Claude Code --stdio--> MCP Server (Node.js) --Socket.io--> Browser Dashboard
| (React 19)
Express HTTP (0.0.0.0)
SQLite (better-sqlite3)
Quick Start
npm install
npm run build:all
npm start
# Dashboard at http://localhost:4800
For development:
npm run dev # Server + Client in parallel
MCP Setup
Add to your Claude Code config (~/.claude.json or .mcp.json):
{
"mcpServers": {
"mcp-dashboard": {
"type": "stdio",
"command": "node",
"args": ["/path/to/mcp-dashboard/dist/index.js"],
"env": { "DASHBOARD_PORT": "4800" }
}
}
}
18 MCP Tools
Workspaces
| Tool | Description |
|---|---|
dashboard_create_workspace |
Create a workspace (name, description, icon) |
dashboard_list_workspaces |
List all workspaces with widget counts |
dashboard_delete_workspace |
Delete a workspace and all its widgets |
Widgets
| Tool | Description |
|---|---|
dashboard_create_widget |
Create a widget in a workspace |
dashboard_update_widget |
Update widget data, title, or config |
dashboard_push_data |
Push incremental data (append rows, points, etc.) |
dashboard_delete_widget |
Remove a widget |
dashboard_list_widgets |
List widgets in a workspace |
Tasks & Reminders
| Tool | Description |
|---|---|
dashboard_add_task |
Add a task to a workspace |
dashboard_list_tasks |
List tasks |
dashboard_complete_task |
Toggle task completion |
dashboard_add_reminder |
Add a reminder with optional due date |
dashboard_list_reminders |
List reminders |
dashboard_complete_reminder |
Toggle reminder completion |
Highlights & Utilities
| Tool | Description |
|---|---|
dashboard_pin_widget |
Pin/unpin widget as highlight |
dashboard_snapshot |
Snapshot workspace state as artifact |
dashboard_notify |
Send toast notification to dashboard |
dashboard_report_activity |
Log MCP activity from other servers |
15 Widget Types
| Type | Description | Streaming |
|---|---|---|
metric_card |
Single KPI with trend and sparkline | Yes |
line_chart |
Time series (Recharts) | Yes |
bar_chart |
Categorical comparison | Yes |
pie_chart |
Distribution donut chart | No |
table |
Sortable, paginated data table | Yes |
markdown |
Rich text with GFM tables | No |
json_viewer |
Collapsible JSON tree | No |
code_block |
Syntax-highlighted code | No |
terminal |
Streaming command output | Yes |
timeline |
Chronological events | Yes |
progress_tracker |
Task progress with overall bar | Yes |
geometry |
SVG art canvas for visual expression | No |
links |
Bookmark collection with auto-icons | Yes |
project |
Project hub: repos, Slack, Notion, workspace link | No |
workspace_ref |
Card referencing another workspace | No |
section |
Full-width divider/title | No |
Data Formats
Each widget type has a specific data schema. The MCP tool description includes all formats. Examples:
// Metric Card
{ value: 158000, label: "Revenue", unit: "$", trend: { direction: "up", percentage: 12.5 }, sparkline: [120, 135, 155, 158] }
// Bar Chart
{ categories: ["North", "South", "East"], series: [{ name: "Q1", values: [42000, 35000, 28000], color: "#60a5fa" }] }
// Line Chart
{ series: [{ name: "Revenue", data: [{ timestamp: "2026-01", value: 45000 }], color: "#60a5fa" }] }
// Table
{ columns: [{ key: "name", label: "Name", sortable: true }], rows: [{ name: "Acme Corp" }] }
// Project
{ name: "Tables", status: "active", repos: [{ name: "tables-sdk", url: "https://...", language: "TypeScript" }], channels: [{ name: "dev", type: "slack" }], notion: [{ title: "Docs", url: "https://...", type: "database" }], workspaceId: "xxx" }
// Geometry (SVG art)
{ svg: "<circle cx='200' cy='200' r='50' fill='#60a5fa'/>", viewBox: "0 0 400 400" }
Features
- Real-time: Socket.io bridge pushes updates instantly to all connected browsers
- Multi-instance: Multiple Claude Code sessions share the same dashboard. First instance owns HTTP, others relay via API
- Drag & resize: react-grid-layout with drag handles and resize
- Highlights: Pin any widget to a featured row at the top
- Tasks & Reminders: Per-workspace todo lists and reminders with due dates
- Archive: Workspaces can be archived and restored
- Expand: Click any widget title to view fullscreen
- Markdown: Full GFM support with expand-to-page
- PDF export: One-click workspace export to PDF
- LAN sharing: Share workspace URLs on local network
- Activity bar: Shows MCP activity stream and overdue reminders
- Workspace references: Meta-dashboards that link to other workspaces
- Dark theme: Futuristic AI aesthetic with neural grid, scan lines, and ambient glow
Tech Stack
Server: Node.js 20, TypeScript, @modelcontextprotocol/sdk, Express 5, Socket.io 4, better-sqlite3, Zod
Client: React 19, Vite 6, TailwindCSS 4.1, Recharts, react-grid-layout, Zustand, Lucide icons, html2canvas-pro, jsPDF
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。