
Visuals MCP
io.github.harrybin由 GitHub 帳號 harrybin 發布v1.6.0更新於 Oct 24, 2026
MCP server for interactive visualizations: tables, images, trees, lists, master-detail
概覽
本機 MCP 伺服器,在 MCP 用戶端中呈現互動式視覺化:表格、圖表、圖片、樹狀、清單與主從檢視。
- 可以做什麼
- 呈現可互動的資料表格,支援排序、篩選、分頁、欄位顯示切換、列選取以及 CSV/PDF 匯出
- 繪製折線、長條、面積、圓餅、散佈與組合圖表,可同畫面多圖,支援圖例、提示框與雙 Y 軸
- 以卡片呈現圖片,含標題、說明、尺寸與檔案大小,來源可為 URL、data URI 或本機路徑
- 以主從版面瀏覽集合,以可展開的樹狀檢視呈現階層資料
- 顯示格式化清單,支援拖放排序、核取方塊、縮圖以及複製與匯出
- 將表格、圖表、樹狀與選取狀態回傳給模型
- 適用情境
- 當助手的回覆以表格、圖表、圖片畫廊、樹狀或可排序清單呈現比純文字更易讀,且用戶端支援 MCP Apps 類介面呈現時適用。主要針對 VS Code 與 GitHub Copilot Chat,其他 MCP 用戶端可透過 npm 套件接入。
- 執行需求
- 以本機 stdio 程序執行;可全域安裝 npm 套件 @harrybin/visuals-mcp,或在用戶端中從 npm 加入,也可使用 VS Code 擴充功能。npm 方式需要 Node.js。未宣告驗證、環境變數或標頭。
安裝
在 SourceWeft 中
- 開啟 儀表板中的 Visuals MCP,將其新增到工作區。
- 為需要使用其工具的對話啟用該服務。
Desktop only,透過 STDIO。 STDIO 服務會啟動本機處理程序,因此需要 SourceWeft 桌面主機。
其他 MCP 客戶端
參照 儲存庫 中的啟動說明。
README
MCP Visuals Server
An MCP (Model Context Protocol) server that provides interactive visualizations for AI agents. Display data in rich, interactive formats including tables with TanStack Table, charts with Recharts, image previews with metadata, master-detail views for browsing collections, tree views for hierarchical data, and customizable lists with drag-and-drop reordering.
Quick Install
[Install with VS Code] °°° [Install with VS Code Insiders]
Features
Table Visualization
- Interactive Table Display: Full-featured data table with TanStack Table v8
- Sorting: Click column headers to sort ascending/descending
- Filtering: Per-column text filters with apply button
- Pagination: Customizable page sizes (5, 10, 20, 50, 100 rows)
- Column Visibility: Toggle which columns are displayed
- Row Selection: Select individual rows or all rows
- Export: Copy as CSV/TSV or export to PDF
- Agent Integration: Table state (selections, filters, sorting) automatically sent back to the LLM via
updateModelContext - Theme Integration: Respects VS Code theme colors and fonts
- Responsive: Works on different screen sizes
- Generic Data Support: Accepts any column structure and data types
Chart Visualization
- Interactive Charts: Render line, bar, area, pie, scatter, and composed charts
- Multiple Charts Per View: Show several charts together in vertical, horizontal, or grid layouts
- Flexible Series Configuration: Configure named series, custom colors, stacked bars/areas, and mixed composed charts
- Rich Chart Features: Optional legends, tooltips, grid lines, and dual Y-axes
- Export Options: Copy chart payloads as JSON or CSV directly from the UI
- Agent Awareness: Active chart selection and hover state can be reported back to the host app
- Theme Integration: Respects VS Code theme colors and fonts
- Responsive: Adapts to narrow panels and wider layouts
Image Preview
- Rich Image Cards: Display images with title, caption, and metadata
- Metadata Display: Show filename, dimensions, and file size
- Flexible Sources: Support URLs and data URIs
- Local File Support: Automatically converts local file paths to data URIs
- Theme Integration: Respects VS Code theme colors and fonts
Master-Detail View
- Flexible Layout: Display a list of items with detailed content panel
- Multiple Content Types: Detail panel supports tables, images, or custom text/HTML
- Reusable Components: Leverages existing table and image visualizations
- Configurable Layout: Choose horizontal (side-by-side) or vertical (stacked) orientation
- State Management: Selection state automatically sent back to LLM
- Rich Master List: Display items with icons, labels, and descriptions
- Theme Integration: Consistent theming across all components
Tree View
- Hierarchical Data Display: Interactive tree structure for nested data
- Expand/Collapse: Click to expand or collapse nodes
- Node Selection: Select individual nodes to highlight them
- Metadata Support: Display optional metadata for each node
- Icons: Add custom icons/emojis to nodes
- Bulk Operations: Expand all or collapse all nodes at once
- Export Options: Copy tree to clipboard, export as HTML, or save as image (PNG)
- Agent Integration: Tree state (expanded nodes, selection) sent back to the LLM
- Theme Integration: Respects VS Code theme colors and fonts
- Use Cases: File systems, org charts, nested categories, JSON/XML structures
List Visualization
- Interactive Lists: Display any type of list with rich formatting
- Drag-and-Drop Reordering: Easily reorder items by dragging
- Checkboxes: Optional checkboxes for task lists and selections
- Image Thumbnails: Show images alongside list items
- Compact Mode: Toggle between comfortable and compact layouts
- Export Options: Copy as plain text, CSV, or JSON
- Individual Item Copy: Quick copy button for each item
- Subtext Support: Secondary text/description for each item
- Metadata: Attach custom metadata to items
- Theme Integration: Respects VS Code theme colors and fonts
Screenshots
Interactive table with sorting, filtering, pagination, and row selection
Interactive chart dashboard with multiple chart types and grid layout
Master-detail view for browsing collections with tables, images, or custom content
Interactive list with drag-and-drop reordering, checkboxes, and image thumbnails
Interactive tree view with expand/collapse, node selection, and export options
Installation
VSCode Extension (Easiest)
Install directly from the VS Code Marketplace:
- Open VS Code
- Go to Extensions (Ctrl+Shift+X / Cmd+Shift+X)
- Search for "Visuals MCP"
- Click Install
The extension will automatically register the MCP server with VS Code and GitHub Copilot Chat.
npmjs (Recommended for other MCP Clients)
In VSCode github Copilot Chat
- click on the tools icon (Configure tools...)
- click on "Add MCP Server"
- click on "Install from npm"
- enter
@harrybin/visuals-mcp - press enter / click "Install"
Using command line:
GitHub Packages
- Authenticate npm with GitHub Packages:
- Install and run:
Local Development
Chart Tool
Use display_chart to render one or more interactive charts in the MCP client.
Supported chart types:
linebarareapiescattercomposed
Repository Metrics Dashboard Example
Generate fresh metrics first:
This writes a snapshot to doc/repo-metrics.json.
Then render a dashboard that uses all chart types (line, bar, area, pie, scatter, composed):
For release-quality docs, always regenerate doc/repo-metrics.json right before updating README examples.
1. Configure MCP Server
For VS Code: This workspace includes MCP configuration in .vscode/settings.json.
To add globally, update your VS Code settings:
For Claude Desktop: See claude_desktop_config.json for example configuration.
License
MIT
Contributing
Contributions welcome! Please open an issue or PR.
Resources
來源:README.md,提交 db17eed
工具
0版本歷史
1- v1.6.0最新Sep 16, 2026
評分
已滿足適用檢查項的 40%
該評分衡量此收錄條目的完整程度,依據註冊中心條目、端點與公開來源計算得出。它不是對伺服器本身的評價,也不表示伺服器安全或目前可用:可連通性是帶日期的檢查結果,評分也不包含安裝量或使用者評價。
- 近 30 天內可連通25 分
請保持端點能夠回應 MCP 握手;SourceWeft 每 30 天探測一次。
- 已宣告認證方式10 分
請在註冊中心條目中宣告憑證請求標頭,或保持端點能夠回應探測,以便判定其認證方式。
- 已列出工具15 分
請讓伺服器在無憑證時回應 tools/list,或發布可在沙箱中無憑證啟動的軟體包。
- 描述必需15 分
- 圖示10 分
- 文件15 分
- 授權條款10 分
請在儲存庫或註冊中心條目中宣告授權條款。