@vprop/mcp

MCP (Model Context Protocol) server for the vProp Public API — let AI assistants create real-estate listing videos on your behalf.

What it does

Connect Claude (or any MCP-compatible AI) to vProp so it can:

• Create and manage property listings
• Upload listing photos
• Create and manage agent profiles
• Generate AI-powered listing videos
• Poll for video completion and return the final URL

Typical flow:

create_listing → upload_listing_image (×10+) → create_agent →
upload_agent_photo → create_project → get_project_status (poll) → get_project

Prerequisites

A vProp API key (VP_…). Contact vProp to get one.

Setup

The MCP server config block is the same across all clients — only the file location differs.

{
  "command": "npx",
  "args": ["-y", "@vprop/mcp"],
  "env": { "VPROP_API_KEY": "VP_your_key_here" }
}

Claude Desktop

Edit ~/Library/Application Support/Claude/claude_desktop_config.json (macOS) or %APPDATA%\Claude\claude_desktop_config.json (Windows):

{
  "mcpServers": {
    "vprop": {
      "command": "npx",
      "args": ["-y", "@vprop/mcp"],
      "env": { "VPROP_API_KEY": "VP_your_key_here" }
    }
  }
}

Restart Claude Desktop.

Claude Code (CLI)

claude mcp add vprop npx -e VPROP_API_KEY=VP_your_key_here -- -y @vprop/mcp

Cursor

Create or edit .cursor/mcp.json in your project root (or ~/.cursor/mcp.json for global):

{
  "mcpServers": {
    "vprop": {
      "command": "npx",
      "args": ["-y", "@vprop/mcp"],
      "env": { "VPROP_API_KEY": "VP_your_key_here" }
    }
  }
}

Restart Cursor or run Cursor: Reload MCP Servers from the command palette.

Windsurf

Edit ~/.codeium/windsurf/mcp_settings.json:

{
  "mcpServers": {
    "vprop": {
      "command": "npx",
      "args": ["-y", "@vprop/mcp"],
      "env": { "VPROP_API_KEY": "VP_your_key_here" }
    }
  }
}

Zed

Edit ~/.config/zed/settings.json:

{
  "context_servers": {
    "vprop": {
      "command": {
        "path": "npx",
        "args": ["-y", "@vprop/mcp"],
        "env": { "VPROP_API_KEY": "VP_your_key_here" }
      }
    }
  }
}

Continue (VS Code / JetBrains)

Edit .continue/config.json in your project (or ~/.continue/config.json globally):

{
  "mcpServers": [
    {
      "name": "vprop",
      "command": "npx",
      "args": ["-y", "@vprop/mcp"],
      "env": { "VPROP_API_KEY": "VP_your_key_here" }
    }
  ]
}

Gemini CLI

Edit ~/.gemini/settings.json:

{
  "mcpServers": {
    "vprop": {
      "command": "npx",
      "args": ["-y", "@vprop/mcp"],
      "env": { "VPROP_API_KEY": "VP_your_key_here" }
    }
  }
}

Other MCP-compatible clients

Any client that follows the MCP specification uses the same stdio transport config:

{
  "command": "npx",
  "args": ["-y", "@vprop/mcp"],
  "env": { "VPROP_API_KEY": "VP_your_key_here" }
}

Note: ChatGPT and Gemini web interfaces do not currently support MCP servers. The tools above are desktop/CLI clients that run the MCP server locally.

Available tools

Voices & Music

| Tool | Description | |------|-------------| | list_voices | List narration voices. Use the id as voice_id. | | list_bgms | List background music tracks. Use the id as bgm_id. |

Listings

| Tool | Description | |------|-------------| | create_listing | Create a property listing (address required) | | get_listing | Get listing details by ID | | list_listings | List all listings (paginated) | | update_listing | Update address, attributes, or metadata | | delete_listing | Delete a listing | | list_listing_images | List uploaded images for a listing | | upload_listing_image | Upload a local JPEG/PNG/WebP to a listing | | delete_listing_image | Remove an image from a listing |

Agents

| Tool | Description | |------|-------------| | create_agent | Create a realtor agent profile | | get_agent | Get agent details by ID | | list_agents | List all agents (paginated) | | update_agent | Update agent info or default voice | | delete_agent | Delete an agent | | upload_agent_photo | Upload a headshot photo (enables avatar feature) |

Projects (video generation)

| Tool | Description | |------|-------------| | create_project | Start video generation (async, returns proj_…) | | get_project_status | Lightweight status check (pending/processing/completed/failed) | | get_project | Full project details including video_url when done | | list_projects | List projects with optional status filter |

Webhooks

| Tool | Description | |------|-------------| | register_webhook | Register an HTTPS URL to receive project events (replaces any existing registration) | | get_webhook | Get the current webhook registration for this API key | | delete_webhook | Deactivate a webhook by its ID | | rotate_webhook_secret | Roll the HMAC signing secret for a webhook |

Usage

| Tool | Description | |------|-------------| | get_usage | Get usage counts and credit balance for this API key |

Example prompts

Generate a video end-to-end:

"Create a listing for 123 Oak Ave, Austin TX 78701. Upload all photos from ~/Desktop/photos/. Create an agent named Jane Doe with email [email protected]. Then generate a video in modern theme and let me know when it's ready."

Check on existing projects:

"Show me all completed projects from the last run."

Try different styles:

"Generate a landscape video with bold theme and no captions for listing lst_AbCd."

Video options

When calling create_project, you can customize:

| Option | Values | Default | |--------|--------|---------| | theme | modern classic bold elegant luxe basic rosewood keystone pixel editorial sketch | modern | | background | black white | theme default | | resolution | portrait (9:16) landscape (16:9) | portrait | | caption | none keyword description | keyword | | language | en es ko zh-Hans | en | | voice_id | ID from list_voices | agent default | | bgm_id | ID from list_bgms | theme default | | background_music | true false | true | | narration_highlights | free text (≤2000 chars) | — | | word_highlight | true false | false | | use_agent_avatar | true false | true | | use_lipsync | true false | false | | display_info | object — per-field booleans for agent card | platform defaults |

Notes

• Image uploads require a local file path — the MCP server runs on your machine and reads files directly. Minimum 10 photos recommended for best results.
• Video generation is async — create_project returns immediately. Poll get_project_status until completed, then call get_project for the video_url.
• Rate limits — project creation is capped at 10/min, 120/hour. Most other endpoints allow 120/min.

Links

• vProp API docs
• API reference (OpenAPI)
• vProp website