@subscribe.dev/mcp

Model Context Protocol (MCP) server for Subscribe.dev - manage projects and deployments directly from Claude Code

Overview

This MCP server enables Claude Code and other MCP-compatible clients to interact with Subscribe.dev directly. Create projects, deploy applications, and manage your Subscribe.dev infrastructure without leaving your AI assistant.

What is MCP?

Model Context Protocol (MCP) is an open standard that allows AI assistants like Claude to securely connect to external tools and data sources. Think of it as "USB-C for AI" - a universal way to extend AI capabilities.

What can this MCP server do?

• Project Management: Create, list, view, and delete Subscribe.dev projects
• Deployments: Deploy HTML/web content to your projects instantly
• File Uploads: Upload files and get public URLs for use with AI models
• AI Model Execution: Run text, image, and video AI models directly from Claude Code
  • Text AI: GPT-4o, GPT-5, Claude, Gemini 3 Pro, and more for chat and completions
  • Image AI: FLUX, Stable Diffusion for image generation and editing
  • Video AI: Sora, SeeDance, Kling Turbo for text-to-video and image-to-video generation
  • Lipsync: Sync audio with video for realistic lip movements
• API Keys: Retrieve project API keys for further development
• Seamless Integration: Works natively with Claude Code and other MCP clients

Installation

Quick Start with Claude Code

The easiest way to use this MCP server is with Claude Code's built-in MCP support:

# Add the MCP server to Claude Code
claude mcp add subscribe-dev

# Or manually add to your Claude Code config

Manual Installation

# Install globally via npm
npm install -g @subscribe.dev/mcp

# Or use npx (no installation required)
npx @subscribe.dev/mcp

Configuration

Required: Platform API Key

You need a Subscribe.dev platform API key to use this MCP server:

1. Visit https://dashboard.subscribe.dev/api-keys
2. Click "Create API Key"
3. Give it a name (e.g., "Claude Code MCP")
4. Choose an expiration period
5. Copy the generated key (starts with sdp_)

Setting up the API Key

Set the environment variable SUBSCRIBE_DEV_API_KEY:

# Option 1: Export in your shell
export SUBSCRIBE_DEV_API_KEY=sdp_your_key_here

# Option 2: Add to your shell profile (~/.bashrc, ~/.zshrc, etc.)
echo 'export SUBSCRIBE_DEV_API_KEY=sdp_your_key_here' >> ~/.zshrc

# Option 3: Use .env file (if your MCP client supports it)
# Create a .env file with:
SUBSCRIBE_DEV_API_KEY=sdp_your_key_here

Claude Code Configuration

Add this to your Claude Code MCP configuration file (~/.claude.json or via claude mcp add):

{
  "mcpServers": {
    "subscribe-dev": {
      "type": "stdio",
      "command": "npx",
      "args": ["-y", "@subscribe.dev/mcp"],
      "env": {
        "SUBSCRIBE_DEV_API_KEY": "sdp_your_key_here"
      }
    }
  }
}

Alternative: If installed globally:

{
  "mcpServers": {
    "subscribe-dev": {
      "type": "stdio",
      "command": "subscribe-dev-mcp",
      "env": {
        "SUBSCRIBE_DEV_API_KEY": "sdp_your_key_here"
      }
    }
  }
}

Available Tools

subscribe_dev_create_project

Create a new Subscribe.dev project.

Parameters:

• name (required): Project name (1-100 characters)
• description (optional): Project description
• type (optional): "demo", "dev", or "production" (default: "dev")
• authCallbackUrl (optional): Callback URL for user authentication
• authCallbackEnabled (optional): Enable auth callback (default: false)

Returns: Project details including the API key ⚠️ Save it - shown only once!

Example:

Create a new Subscribe.dev project called "My AI App"

subscribe_dev_list_projects

List all your Subscribe.dev projects with pagination.

Parameters:

• limit (optional): Number of projects to return (1-100, default: 10)
• offset (optional): Pagination offset (default: 0)

Returns: Array of projects and pagination metadata

Example:

List my Subscribe.dev projects

subscribe_dev_get_project

Get detailed information about a specific project.

Parameters:

• projectId (required): Project ID (UUID format)

Returns: Complete project details

Example:

Get details for project d789e442-5b0b-4837-8a03-95171a05d74f

subscribe_dev_get_project_api_key

Retrieve the API key for a project (for making API requests).

Parameters:

• projectId (required): Project ID (UUID format)

Returns: The project's API key

Example:

Get the API key for project d789e442-5b0b-4837-8a03-95171a05d74f

subscribe_dev_delete_project

Delete a project permanently.

Parameters:

• projectId (required): Project ID to delete (UUID format)

Returns: Confirmation message

Example:

Delete project d789e442-5b0b-4837-8a03-95171a05d74f

subscribe_dev_deploy_content

Deploy HTML or file content to a project. Your content goes live instantly!

Parameters:

• projectId (required): Project ID to deploy to (UUID format)
• content (required): HTML/file content to deploy
• isBase64 (optional): Whether content is base64 encoded (default: false)

Returns: Deployment ID and live URL

Example:

Deploy this HTML to project d789e442-5b0b-4837-8a03-95171a05d74f:
<!DOCTYPE html>
<html>
<head><title>Hello World</title></head>
<body><h1>Hello from Claude Code!</h1></body>
</html>

upload_file

Upload a file (image, video, audio) and get a public CloudFront URL. Perfect for uploading reference images to use with AI models.

Parameters:

• filePath (required): Absolute path to the file to upload

Supported file types:

• Images: jpg, jpeg, png, webp, gif
• Videos: mp4, webm, mov
• Audio: mp3, wav

Returns: Public CloudFront URL

Example:

Upload the file at /Users/alice/photo.jpg

Use with AI models:

1. Upload a reference image
2. Use the returned URL with run_image_ai or run_video_ai

run_text_ai

Execute text AI models (GPT, Claude, Gemini, etc.) and bill usage to your demo project.

Parameters:

• model (required): Model identifier (e.g., "openai/gpt-4o", "openai/gpt-5", "google/gemini-3-pro-preview", "anthropic/claude-3.5-sonnet")
• input (required): Object containing:
  • prompt (optional): Text prompt for completion

Returns: AI-generated text response with metrics

Examples:

Generate a haiku about coding using GPT-4o

Generate a haiku about coding using GPT-5

Generate a haiku about coding using Gemini 3 Pro

run_image_ai

Generate images using AI models (FLUX, Stable Diffusion, etc.) and bill usage to your demo project.

Parameters:

• model (required): Model identifier (e.g., "black-forest-labs/flux-schnell", "google/nano-banana-pro")
• input (required): Object containing:
  • prompt (required): Description of the image to generate
  • aspect_ratio (optional): Image aspect ratio (e.g., "16:9", "1:1")
  • output_format (optional): Output format ("jpg", "png", "webp")
  • image_input (optional): Array of reference image URLs for nano-banana-pro
  • input_image (optional): Single reference image URL for flux-kontext-max
  • remove_background (optional): Remove background (returns PNG with transparency)

Returns: Image URL(s) with metrics

Example:

Generate an image of a sunset over mountains using flux-schnell

run_video_ai

Generate or process videos using AI models (Sora, SeeDance, Kling Turbo, Lipsync, etc.) and bill usage to your demo project.

Parameters:

• model (required): Model identifier
  • Text-to-video: "openai/sora-2", "openai/sora-2-pro", "bytedance/seedance-1-lite", "bytedance/seedance-1-pro", "kwaivgi/kling-v2.5-turbo-pro"
  • Lipsync: "sync/lipsync-2"
• input (required): Object containing model-specific parameters

Text-to-Video Models:

• prompt (required): Description of the video to generate
• image (optional): Start frame image URL (seedance models)
• start_image (optional): First frame image URL (kling models)
• input_reference (optional): First frame image URL (sora models)
• aspect_ratio (optional): Video aspect ratio
  • seedance: "16:9", "9:16", "1:1", "4:3", "3:4", "21:9", "9:21" (default: "16:9")
  • sora: "portrait" (720x1280), "landscape" (1280x720) (default: "portrait")
  • kling: "16:9", "9:16", "1:1" (default: "16:9", ignored if start_image provided)
• seconds (optional): Duration in seconds (sora: 4/8/12)
• duration (optional): Duration in seconds (seedance: 2-12, kling: 5 or 10, default: 5)
• resolution (optional): Video resolution (model-specific)
• negative_prompt (optional): Things to avoid in the video (kling)
• camera_fixed (optional): Lock camera position (seedance)
• last_frame_image (optional): End frame for looping videos (seedance)

Lipsync Model (sync/lipsync-2):

• video (required): Input video URL or base64 to apply lip synchronization
• audio (required): Audio URL or base64 to synchronize with the video
• sync_mode (optional): Synchronization mode
  • "loop": Loop audio to match video length (default)
  • "bounce": Reverse audio playback when looping
  • "cut_off": Cut off audio at video end
  • "silence": Add silence to match lengths
  • "remap": Remap audio timing to match video
• temperature (optional): Generation randomness (0.0-1.0, default: 0.5)

Returns: Video URL with metrics

Examples:

Generate a 5-second video of a cat playing with yarn using seedance-1-pro

Generate a 10-second cinematic video of a surfer riding a wave at sunset using kling-v2.5-turbo-pro

Generate a video from this image https://example.com/sunset.jpg showing the scene transitioning from day to night using kling-v2.5-turbo-pro

Sync this audio https://example.com/speech.mp3 with this video https://example.com/person.mp4 using lipsync-2

Pricing:

• Kling v2.5 Turbo Pro: $0.07 per second of video output
• Lipsync models: $0.0525 per second of video processed

Usage Examples

In Claude Code

Once configured, you can use natural language to interact with Subscribe.dev:

You: Create a new dev project called "Todo App"

Claude: I'll create a new dev project for you.
[Creates project using subscribe_dev_create_project]
✓ Project created! Here are the details:
- ID: abc123...
- Name: Todo App
- API Key: pub_xyz... (Save this securely!)
- Type: dev

You: Deploy a simple HTML page to that project

Claude: I'll deploy some HTML content to your Todo App project.
[Deploys using subscribe_dev_deploy_content]
✓ Deployment successful!
Your site is live at: https://abc123.apps.subscribe.dev

You: List all my projects

Claude: Here are your Subscribe.dev projects:
[Lists using subscribe_dev_list_projects]
1. Todo App (dev)
2. Portfolio Site (production)
3. Test Project (demo)
...

Development

Building from Source

# Clone the repository
git clone https://github.com/volter-ai/remetricate.git
cd remetricate/packages/mcp

# Install dependencies
npm install

# Build
npm run build

# Run locally
SUBSCRIBE_DEV_API_KEY=sdp_your_key_here node dist/index.js

Project Structure

packages/mcp/
├── src/
│   └── index.ts          # Main MCP server implementation
├── dist/                 # Compiled output
├── package.json
├── tsconfig.json
└── README.md

Troubleshooting

"SUBSCRIBE_DEV_API_KEY environment variable is required"

Make sure you've set the SUBSCRIBE_DEV_API_KEY environment variable with your platform API key. Get one from https://dashboard.subscribe.dev/api-keys

"Invalid API key format"

Platform API keys must start with sdp_. Make sure you're using a platform API key, not a project API key (which starts with pub_).

"Module not found"

If you installed globally, make sure subscribe-dev-mcp is in your PATH. Try using npx instead:

{
  "command": "npx",
  "args": ["-y", "@subscribe.dev/mcp"]
}

MCP server not appearing in Claude Code

1. Check your configuration file syntax
2. Restart Claude Code
3. Run claude mcp list to see registered servers
4. Check logs with claude mcp get subscribe-dev

Related Packages

• @subscribe.dev/sdk - TypeScript SDK for Subscribe.dev
• @subscribe.dev/react - React hooks for Subscribe.dev
• @subscribe.dev/client - Core client library

Resources

• Subscribe.dev Documentation
• Subscribe.dev Dashboard
• Model Context Protocol Docs
• Claude Code Documentation

Support

• Issues: GitHub Issues
• Discussions: GitHub Discussions
• Email: [email protected]

License

MIT © Volter AI

Built with Model Context Protocol | Powered by Subscribe.dev