keynote-mcp

keynote-mcp

Local macOS MCP server that reads, edits, and exports Apple Keynote presentations via JXA. Enables Claude to visually analyze slides, apply design changes, and iterate on presentations directly.

Category
访问服务器

README

keynote-mcp

Local macOS MCP server that reads, edits, and exports Apple Keynote presentations via JXA (JavaScript for Automation). Enables Claude to visually analyze slides, apply design changes, and iterate on presentations directly — without any third-party tools.

Requirements

  • macOS with Apple Keynote installed
  • Node.js 20+
  • Keynote must be running with at least one document open

Installation

git clone https://github.com/tszaks/keynote-mcp
cd keynote-mcp
npm install && npm run build

Add to ~/.mcp.json:

{
  "mcpServers": {
    "keynote-mcp": {
      "type": "stdio",
      "command": "/opt/homebrew/bin/node",
      "args": ["/path/to/keynote-mcp/dist/index.js"]
    }
  }
}

Restart Claude Code. Tools appear as mcp__keynote-mcp__keynote_*.


Tools Reference

All tools accept doc_index (0-based, defaults to 0 = frontmost document). Elements are addressed by slide_index (0-based) + element_type (textItem, imageItem, shapeItem) + element_index (0-based within that collection). Colors are [r, g, b] with values 0–255. Slide coordinates are in points; standard Keynote slide is 1920×1080 pts.

Discovery

keynote_list_presentations

List all open Keynote documents.

Input: {}
Output: [{ index, name, slideCount, modified, filePath }]

keynote_get_presentation_info

Get slide count and available master slide names. Call this first before redesigning — you need master names to call keynote_apply_master.

Input: { doc_index? }
Output: { name, slideCount, masterSlides: string[], filePath, modified }

keynote_get_slide

Get the full element tree for one slide: all text boxes, images, and shapes with position, size, font, font size, and color.

Input: { doc_index?, slide_index }
Output: {
  slideIndex, masterSlide, skipped, elementCount,
  elements: [{
    type: "textItem" | "imageItem" | "shapeItem",
    index,
    text?,         // textItem only
    font?,         // textItem only
    fontSize?,     // textItem only
    color?,        // textItem only — [r,g,b] 0-255
    fillColor?,    // shapeItem only — [r,g,b] 0-255
    fileName?,     // imageItem only
    position: { x, y },
    size: { w, h }
  }]
}

Export (Visual Analysis)

keynote_export_slide_image

Export a single slide as PNG. Returns the file path — use the Read tool on that path to see the slide visually.

Input: { doc_index?, slide_index }
Output: { filePath, slideIndex, exportDir }

keynote_export_all_slides

Export all slides as PNG to ~/Library/Caches/keynote-mcp/<doc>/. Returns array of file paths.

Input: { doc_index? }
Output: { exportDir, slideCount, files: string[] }

keynote_get_design_snapshot

Best starting point for redesign. One call that exports all slides as PNG AND returns the full element tree for every slide.

Input: { doc_index? }
Output: {
  presentationName, slideCount, exportDir,
  slideImages: string[],   // PNG file paths — Read these for visual analysis
  slides: SlideData[]      // full element trees
}

Text & Font

keynote_set_text

Replace the text content of a text element. Preserves existing font styling.

Input: { doc_index?, slide_index, element_type: "textItem", element_index, text }

keynote_set_font

Set font family, size, and/or color on a text element. Omit any field to leave it unchanged.

Input: {
  doc_index?, slide_index,
  element_type: "textItem", element_index,
  font_name?,   // e.g. "SF Pro Display", "Helvetica Neue", "Didot"
  font_size?,   // points
  color?        // [r,g,b] 0-255
}

Layout

keynote_set_element_position

Move an element. x = distance from left edge, y = distance from top edge, both in points.

Input: { doc_index?, slide_index, element_type, element_index, x, y }

keynote_set_element_size

Resize an element.

Input: { doc_index?, slide_index, element_type, element_index, width, height }

Color & Fill

keynote_set_slide_background

Set slide background to a solid color. May fail if the slide's master has a locked background — apply a blank master first with keynote_apply_master.

Input: { doc_index?, slide_index, color: [r,g,b] }

keynote_set_element_fill_color

Set the fill color of a shape element.

Input: { doc_index?, slide_index, element_type: "shapeItem", element_index, color: [r,g,b] }

Add & Delete Elements

keynote_add_text_box

Add a new text box to a slide.

Input: { doc_index?, slide_index, text?, x?, y?, width?, height? }
Output: { added, slideIndex, newElementIndex }

keynote_add_shape

Add a rectangle or oval shape, with optional fill color.

Input: { doc_index?, slide_index, x?, y?, width?, height?, fill_color?: [r,g,b] }
Output: { added, slideIndex, newElementIndex }

keynote_delete_element

Remove an element from a slide. Keynote undo (Cmd+Z) still works in the app.

Input: { doc_index?, slide_index, element_type, element_index }

Slides

keynote_apply_master

Apply a master slide to one slide or all slides. Get valid master names from keynote_get_presentation_info.

Input: { doc_index?, master_name, slide_index? }
// Omit slide_index to apply to all slides

keynote_add_slide

Append a new blank slide at the end.

Input: { doc_index? }
Output: { added, newSlideIndex, slideCount }

keynote_delete_slide

Delete a slide.

Input: { doc_index?, slide_index }

keynote_duplicate_slide

Duplicate a slide. The copy is inserted immediately after the original.

Input: { doc_index?, slide_index }

Workflow Patterns

Visual Design Review

1. keynote_get_design_snapshot          → get all slide images + element trees
2. Read(slideImages[0])                 → see slide 0 visually
3. Read(slideImages[1])                 → see slide 1 visually
   ...
4. Identify issues: inconsistent fonts, low contrast, crowded layouts
5. Apply fixes with set_font / set_element_position / set_slide_background
6. keynote_export_slide_image           → verify the result visually

Enforce Typography Consistency

1. keynote_get_slide for each slide     → collect all font names and sizes
2. Identify the dominant font family
3. keynote_set_font on outliers to match

Redesign a Single Slide

1. keynote_export_slide_image           → see current state
2. keynote_get_slide                    → get element positions and sizes
3. Apply repositions, resizes, font changes
4. keynote_export_slide_image           → verify result

Known Limitations

  • Export requires a saved file — if filePath is empty in list_presentations, save the document first (Cmd+S). Export will fail on unsaved presentations.
  • Element positions return 0,0 for some Keynote layouts — this is a JXA limitation with certain master/layout combinations. Text content and font data are unaffected.
  • Background color changes may be blocked by master slide locks. Use keynote_apply_master with a blank master first.
  • No animation access — JXA does not expose Keynote transition or build animation properties.
  • Image content is not readableimageItem elements return file name and dimensions only, not pixel data.

Architecture

TypeScript MCP server using the @modelcontextprotocol/sdk. All Keynote interaction goes through JXA (JavaScript for Automation) via osascript -l JavaScript. Operations and payloads are passed as environment variables; results return as a JSON envelope { ok: boolean, result? | error? } on stdout.

Follows the same pattern as reminders-mcp, safari-mcp, and other local macOS MCP servers in the same ecosystem.

推荐服务器

Baidu Map

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。

官方
精选
JavaScript
Playwright MCP Server

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。

官方
精选
TypeScript
Magic Component Platform (MCP)

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。

官方
精选
本地
TypeScript
Audiense Insights MCP Server

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。

官方
精选
本地
TypeScript
VeyraX

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。

官方
精选
本地
graphlit-mcp-server

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。

官方
精选
TypeScript
Kagi MCP Server

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。

官方
精选
Python
e2b-mcp-server

e2b-mcp-server

使用 MCP 通过 e2b 运行代码。

官方
精选
Neon MCP Server

Neon MCP Server

用于与 Neon 管理 API 和数据库交互的 MCP 服务器

官方
精选
Exa MCP Server

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。

官方
精选