# SlideMaster Agent

> AI teaching design agent — conversational AI that designs complete course presentations with learning activities, energy management, and professional slide rendering.

## Categories

- AI instructional design tool
- AI course creation
- AI presentation maker for educators
- AI slide generator
- Corporate training content creation
- Teaching video production
- Learning experience design
- AI e-learning authoring
- PPT to video converter
- Voice cloning TTS

SlideMaster Agent (agent.slidemaster.tw) is an AI-powered teaching design assistant built for instructors, corporate trainers, and HR teams. Through natural conversation, it designs full course presentations: generates outlines from topics or YouTube videos, creates structured YAML slide specs, renders slides with Gemini AI, writes narration scripts, generates TTS audio (with voice cloning), and compiles final MP4 videos. It also designs learning activities (polls, discussions, quizzes) with energy curve management for engaging live sessions.

## Product

- [SlideMaster Agent](https://agent.slidemaster.tw): Conversational AI for teaching design — chat to create course presentations
- [SlideMaster Platform](https://slidemaster.tw): Main website with project dashboard, settings, and file management
- [API Documentation](https://slidemaster.tw/api-docs): Full interactive API reference with 20 endpoints
- [API Playground](https://slidemaster.tw/api-playground): Test API endpoints interactively in the browser
- [OpenAPI Spec](https://api.slidemaster.tw/api/v1/public/openapi.json): Machine-readable API schema (JSON)

## API Reference

- [Complete API docs for LLMs](https://slidemaster.tw/llms-full.txt): Full endpoint reference with parameters, responses, and code examples
- [SlideMaster Platform llms.txt](https://slidemaster.tw/llms.txt): Platform-level documentation index

### Authentication

- Base URL: `https://api.slidemaster.tw/api/v1/public`
- Auth header: `X-API-Key: sm_live_...` or `Authorization: Bearer sm_live_...`
- [Get API key](https://slidemaster.tw/settings/api-keys): Create and manage API keys

### Content Creation Endpoints

- POST /outlines/generate: Generate a structured slide outline from a topic using AI
- POST /projects: Create a new project
- POST /upload/init: Get a signed upload URL for PDF/PPTX file upload
- POST /upload/complete: Confirm file upload and trigger server-side parsing
- POST /slides/render: Generate AI slide images with Gemini (supports 1K/2K/4K resolution)

### Content Processing Endpoints

- POST /scripts/generate: Generate narration script for a single slide
- POST /scripts/batch-generate: Batch generate narration scripts for all slides in a project
- POST /tts/generate: Generate text-to-speech audio (Azure, Gemini, CosyVoice providers)
- POST /video/generate: Compose final MP4 video from slides and audio

### Management Endpoints

- GET /projects: List all projects for the authenticated user
- GET /projects/{id}: Retrieve project details and status
- PATCH /projects/{id}: Update project title, TTS settings, or voice
- DELETE /projects/{id}: Permanently delete a project and all its assets
- GET /projects/{id}/slides: List all slides with image/audio URLs
- PATCH /slides/{id}: Update slide title or narration script
- DELETE /slides/{id}: Delete a single slide

### Query & Export Endpoints

- GET /projects/{id}/status: Poll processing status and progress (pending → processing → completed)
- GET /projects/{id}/export: Get download URLs for video, slide images, and audio files
- GET /voices: List all available TTS voices across providers

### Automation Endpoint

- POST /pipelines/topic-to-video: One-call pipeline — topic → outline → slides → scripts → TTS → video

## MCP Server (Model Context Protocol)

Connect AI assistants (Claude, Cursor, ChatGPT) to SlideMaster:

- MCP endpoint (Streamable HTTP): `https://mcp.slidemaster.tw/mcp`
- NPM package (stdio): `npx -y @slidemaster/mcp-server`
- Smithery: `https://mcp--slidemaster.run.tools`
- Auth: OAuth 2.1 (Claude.ai) or API Key via env var

### MCP Tools (20 total)

- generate_outline: Plan a presentation from a topic description
- create_project: Create a new empty project, returns project_id
- upload_init: Get a pre-signed URL to upload PPTX/PDF files
- upload_complete: Trigger server-side processing after file upload
- render_slides: Generate AI slide images for a project
- generate_script: Generate narration script for one slide
- batch_generate_scripts: Generate narration scripts for all slides in a project
- generate_tts: Generate text-to-speech audio for slides
- generate_video: Compile slides + audio into final MP4 video
- topic_to_video: One-click: topic → complete narrated video
- list_projects: Browse existing projects with pagination
- get_project: Retrieve detailed metadata and status for a project
- update_project: Change project title, TTS provider, voice, or language
- delete_project: Permanently delete a project and all assets
- list_slides: List all slides in a project with titles, scripts, and render status
- update_slide: Edit a slide's title or narration script text
- delete_slide: Remove a single slide from a project
- check_status: Poll processing progress after render, TTS, or video generation
- export_project: Get download URLs for all generated assets
- list_voices: Browse available TTS voices across all providers

### Connect from Claude.ai

1. Open claude.ai → Settings → Integrations
2. Click "Add more integrations" → paste: `https://mcp.slidemaster.tw/mcp`
3. Authorize with Google login

### Connect from Claude Desktop / Cursor

```json
{
  "mcpServers": {
    "slidemaster": {
      "command": "npx",
      "args": ["-y", "@slidemaster/mcp-server"],
      "env": { "SLIDEMASTER_API_KEY": "sm_live_YOUR_KEY" }
    }
  }
}
```

## Key Features

- Multi-language support: Traditional Chinese (zh-TW), Simplified Chinese (zh-CN), English (en), Japanese (ja)
- Brand asset overlay: logo, footer, QR code composited on user-specified slides (last page, first+last, all, or custom page numbers)
- Style analysis: upload a reference slide image to extract and replicate its visual style
- Mixed rendering: combine original PowerPoint slides with AI-redesigned slides in one project
- Voice cloning: clone your voice for personalized TTS narration
- Course series planning: design multi-chapter, multi-session curriculum
- Learning activities: design interactive polls, quizzes, discussions, case studies with energy curve planning
- Video export: MP4 with subtitles (SRT/VTT), intro/outro/midroll video insertion

## Guides

- [Agent 對話生簡報入門指南](https://slidemaster.tw/guides/agent-slide-generation-guide): How to use conversational AI to generate a complete presentation
- [AI 體驗藍圖教學設計](https://slidemaster.tw/guides/experience-blueprint-guide): Design an entire lesson's teaching flow with one conversation
- [MCP 整合指南](https://slidemaster.tw/guides/mcp-integration-guide): Connect Claude Desktop and ChatGPT to create presentations via MCP
- [精準控時指南](https://slidemaster.tw/guides/precise-timing-guide): AI-powered script timing estimation so you never go over time
- [AI 投影片生成與 YAML 規格](https://slidemaster.tw/guides/ai-slide-generation-yaml-notebooklm): From visual reverse-engineering to YAML spec — creating slides with soul
- [聲音復刻 Voice Cloning 指南](https://slidemaster.tw/guides/voice-cloning-guide): Create personalized AI narration with your own cloned voice
- [片頭片尾與中插影片](https://slidemaster.tw/guides/intro-outro-midroll-guide): Add intro, outro, and midroll videos to your AI presentations
- [Canva 動畫匯入指南](https://slidemaster.tw/guides/canva-animation-import-guide): Import Canva animations with the critical 5-second setting
- [PDF/PPTX 壓縮工具](https://slidemaster.tw/guides/pdf-pptx-compression-guide): One-click file compression without quality loss
- [企業內訓影片製作流程](https://slidemaster.tw/guides/corporate-video-workflow): Save 80% production time with AI for corporate training videos
- [企業內訓教材製作指南](https://slidemaster.tw/guides/internal-training-content-creation): End-to-end guide for creating corporate training materials from scratch
- [企業內訓常見踩雷](https://slidemaster.tw/guides/training-pitfalls-to-avoid): 5 common corporate training content creation mistakes to avoid
- [傳統錄課 vs AI 轉型](https://slidemaster.tw/guides/traditional-recording-vs-ai-video): Why instructors fear recording, and how AI solves the 5 major pain points
- [PPT 變教學影片技巧](https://slidemaster.tw/guides/ppt-to-video-tips): 3 techniques to turn PowerPoint into engaging teaching videos
- [ADDIE 框架 AI 快速應用](https://slidemaster.tw/guides/addie-framework-ai): Apply the ADDIE instructional design framework in the AI era
- [柯氏四級評估模型](https://slidemaster.tw/guides/kirkpatrick-model-training): Scientifically measure corporate training effectiveness with Kirkpatrick Model
- [教學影片字幕指南](https://slidemaster.tw/guides/video-subtitles-guide): Best practices for automated subtitle generation
- [SRT vs VTT 字幕格式比較](https://slidemaster.tw/guides/srt-vs-vtt-guide): Complete comparison of subtitle formats for teaching videos
- [影片章節與索引設計](https://slidemaster.tw/guides/video-chapters-index): Improve knowledge retrieval with chapter and index design
- [講師錄音 vs AI 配音](https://slidemaster.tw/guides/audio-quality-ai-voice): Why AI voice is a more stable choice for teaching content
- [EverCam 匯出指南](https://slidemaster.tw/guides/what-is-evercam-export): Integrate with enterprise LMS via EverCam export
- [線上內訓平台上架檢核表](https://slidemaster.tw/guides/lms-upload-checklist): Ensure the best learner experience when publishing to your LMS