Presentation Agent MCP
Click on "Install Server".
Wait a few minutes for the server to deploy. Once ready, it will show a "Started" state.
In the chat, type
@followed by the MCP server name and your instructions, e.g., "@Presentation Agent MCPcreate a deck about microservices architecture"
That's it! The server will respond to your query, and you can continue using it as needed.
Here is a step-by-step guide with screenshots.
Presentation Agent MCP
一个基于 TypeScript 的 Presentation MCP Server —— 通过结构化 Deck JSON 和 Layout Engine 自动生成技术汇报 PPTX。
What it does
通过 MCP tools 提供以下能力:
create_deck: 根据主题自动生成结构化 Deck JSON(6-10 页)
get_deck: 读取已生成的 Deck JSON
review_deck: 审查 PPT 内容结构(标题过长、bullet 过多、缺页等)
update_slide: 对单页进行增量修改
export_pptx: 导出真实
.pptx文件create_storyline: 生成故事线
create_style_profile: 获取风格配置
Related MCP server: marp-agent-mcp
Architecture
Client (Codex/ChatGPT/Claude)
│ MCP
▼
Presentation MCP Server (stdio)
│
▼
Presentation Runtime
├── Storyline Planner → 生成故事线
├── Deck Generator → 生成结构化 Deck JSON
├── Layout Engine → 布局计算
├── Review Engine → 内容审查
├── Incremental Editor → 单页修改
└── PptxExporter → PPTX 导出核心约束:所有内容必须先进入 Deck JSON,所有布局必须经过 Layout Engine,LLM 不允许直接写 PPTX。
Install
cd presentation-agent-mcp
npm installRun locally
# Start the MCP server (stdio mode)
npm startFor development with auto-reload:
npm run devTest
npm testTest coverage
Schema tests: Deck/Slide schema validation
Layout tests: All 6 slide types render valid elements within bounds
Review tests: 10+ static check rules
Export tests: PPTX buffer generation with real PptxGenJS
MCP tools
Tool | Description | Input |
| Get style profile |
|
| Generate storyline |
|
| Create deck JSON |
|
| Read deck JSON |
|
| Review deck structure |
|
| Update a single slide |
|
| Export to PPTX |
|
Example workflow
Full closed loop
1. create_deck
2. review_deck
3. export_pptx
4. update_slide
5. export_pptx againVia MCP Inspector
npx @modelcontextprotocol/inspector node --loader tsx src/index.tsExample create_deck input
{
"topic": "Agentic AI 时代端侧 OS 架构演进",
"audience": "操作系统架构师和技术管理层",
"purpose": "说明端侧负载变化,并提出 AgentOS Kernel 架构演进方向",
"slide_count": 8,
"style_id": "allen_huawei_tech"
}Built-in styles
default: Clean default presentation style
allen_huawei_tech: 面向技术战略、架构演进的技术汇报风格
16:9 宽屏(13.333 x 7.5 inch)
主色调: #A80000
字体: Arial
支持 title/agenda/insight/comparison/architecture/summary
Slide types
title: 封面页
agenda: 议程页(最多 6 项)
insight: 洞察页(关键观点 + 证据卡片)
comparison: 对比页(左右两栏)
architecture: 架构页(分层架构图)
summary: 总结页(3 Takeaways + 下一步行动)
Current limitations (V0.1)
No real LLM: 使用确定性规则生成内容,不接外部 LLM
No Deep Research: V0.1 不包含联网研究能力
No UI: 纯 MCP Server,无 Web 界面
Basic layouts: 固定坐标布局,非 constraint-based
No image support: V0.1 不包含图片搜索和渲染
No PDF export
Local storage only: 使用本地 filesystem 存储
No complex animations
Roadmap
V0.2
接入真实 LLM
Deck Patch 机制
更强的 slide-level incremental editing
增加 roadmap、timeline、case_study 页面类型
Remote MCP 部署
V0.3
接入 Deep Research
企业知识库集成
Web Console + PPT 预览
图标库和 SVG 架构图生成
V1.0
ChatGPT App 化
多用户 / 权限 / 模板市场
OpenDesk Presentation Agent 集成
Development
npm run typecheck # TypeScript type check
npm test # Run tests
npm run dev # Start with tsx watchMaintenance
Resources
Unclaimed servers have limited discoverability.
Looking for Admin?
If you are the server author, to access and configure the admin panel.
Latest Blog Posts
- Your AI Chatbot Just Exposed Your CEO's Salary to an InternBy Om-Shree-0709 on .Agent IdentityMCP SecurityOAuth Delegation
- Why MCP Servers Need Execution Sandboxing (And Why Your Current Stack Isn't Enough)By Om-Shree-0709 on .Agentic AiPrompt InjectionWebAssembly
MCP directory API
We provide all the information about MCP servers via our MCP API.
curl -X GET 'https://glama.ai/api/mcp/v1/servers/gaowanlong/presentation-agent-mcp'
If you have feedback or need assistance with the MCP directory API, please join our Discord server