MCP (Model Context Protocol) Server for the SlideMaster Public API. This server exposes 20+ tools that let any MCP-compatible AI assistant create AI-powered presentation videos from a simple topic.

1. Go to slidemaster.tw and sign in with Google
2. Click your avatar (top right) → Settings
3. Navigate to API Keys
4. Click Generate Key — copy the sm_live_... key

Claude Desktop — add to claude_desktop_config.json:

• macOS: ~/Library/Application Support/Claude/claude_desktop_config.json
• Windows: %APPDATA%\Claude\claude_desktop_config.json

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

Restart your AI client. Then just say: "Make a 5-slide presentation about [your topic]"

"Make slides about machine learning"
         │
         ▼
┌─────────────────┐
│ generate_outline │ ← Generates outline + auto-creates project
│                  │   Returns: project_id, slides array
└────────┬────────┘
         ▼
┌─────────────────┐
│  render_slides   │ ← Generates AI images for each slide
│                  │   Costs: 140 credits/slide (2K)
│                  │   Poll check_status until status="parsed"
└────────┬────────┘
         ▼
┌───────────────────────┐
│ batch_generate_scripts │ ← Generates narration scripts (optional)
└────────┬──────────────┘
         ▼
┌─────────────────┐
│  generate_tts    │ ← Text-to-speech audio (optional)
│                  │   Costs: 16 credits/slide
└────────┬────────┘
         ▼
┌─────────────────┐
│ generate_video   │ ← Compile final MP4 (optional)
│                  │   Costs: 80 credits/project
│                  │   Poll check_status until status="video_completed"
└────────┬────────┘
         ▼
┌─────────────────┐
│  export_pdf      │ ← Download as PDF (pdf_url field)
│  export_pptx     │ ← Download as PPTX (pptx_url field)
│  export_project  │ ← Get all download URLs
└─────────────────┘

Shortcut: Use topic_to_video to run the entire pipeline in one call.

Tip: generate_outline automatically creates a project — you do NOT need to call create_project separately.

Quick math: 2,000 credits ≈ 14 slides at 2K (slides only). A full 25-slide course with video ≈ 3,980 credits.

Call get_credit_balance before large projects to check remaining credits.

When polling check_status, the status field means:

To apply a visual style, use analyze_style_image with a publicly accessible image URL (e.g. Imgur, Cloudinary). Base64 is not supported. The returned style_guide object can be passed to render_slides.

User: Create a 5-slide presentation about renewable energy in Taiwan
Claude: I'll create that for you.
[Calls generate_outline with topic="renewable energy in Taiwan", slides_count=5]
→ Returns project_id and slides array
[Calls render_slides with the project_id and slides]
→ Rendering started, polling...
[Calls check_status — status: "parsed", all slides rendered]
[Calls export_pdf]
→ Returns pdf_url
Here's your presentation! Download the PDF: [link]

npm install
npm run build
SLIDEMASTER_API_KEY=your-key npm start

Full API docs: slidemaster.tw/api-docs

MIT