Roomvana MCP Server
# Roomvana MCP server
An [MCP](https://modelcontextprotocol.io) server for **[Roomvana](https://roomvana.ai)** — AI interior design. It lets an AI assistant browse the room types and design styles Roomvana supports and hand the user a studio link with their choices pre-selected. Upload a photo of any room and get a realistic redesign in about 30 seconds at **[roomvana.ai](https://roomvana.ai)**.
Read-only and no sign-in required: this server never uploads photos or spends credits. The actual redesign happens on the website.
## Tools
| Tool | What it does |
| --- | --- |
| `list_design_styles` | Lists the design styles Roomvana can render (modern, japandi, industrial, coastal, …), each with a short description. |
| `list_room_types` | Lists the rooms and outdoor spaces Roomvana can redesign — interior rooms, house exterior, and garden/patio/backyard. |
| `design_a_room` | Returns a [roomvana.ai](https://roomvana.ai) studio link with the chosen room and style pre-selected; validates both against Roomvana's live catalog. |
## Install
Requires Node.js 18+.
```json
{
"mcpServers": {
"roomvana": {
"command": "npx",
"args": ["-y", "roomvana-mcp"]
}
}
}
```
Add that to your MCP client config (Claude Desktop: `claude_desktop_config.json`; other clients: their MCP config file), then restart the client.
### From source
```bash
npm install
npm run build
npm start
```
## Example
> "I want to redo my bedroom in a japandi style."
The assistant calls `design_a_room` with `room=bedroom`, `style=japandi` and returns:
```
https://roomvana.ai/design?room=bedroom&style=japandi
```
Open it, upload a photo of the room, and generate a redesign. Your walls, windows and layout stay put — only the look changes.
## How it works
The list of supported rooms and styles is fetched live from Roomvana's public catalog endpoint (`https://api.roomvana.co/options`), so the server never offers something Roomvana can't render. If that endpoint is unreachable, it falls back to a bundled snapshot.
## Links
- Website: https://roomvana.ai
- Design studio: https://roomvana.ai/design
## License
MIT
TDQS
Scored across 3 tools
Each tool targets a distinct, non-overlapping purpose: two lookup helpers for separate dimensions (styles vs. rooms) and one action tool that composes them. There is no realistic way to confuse them, and the descriptions explicitly cross-reference each other to reinforce the intended workflow.
All three tools use a clean snake_case verb_noun pattern: list_design_styles, list_room_types, and design_a_room. The two listing tools are perfectly parallel, and the action tool follows the same verb-first convention.
Three tools is reasonable for a narrowly scoped link-generation flow: two enumeration helpers plus one generator. It sits at the lean end of appropriate, but each tool clearly earns its place with no redundancy.
The surface covers the core flow of discovering valid styles/rooms and producing a pre-configured design link. Minor gaps exist (e.g. no way to check the status or result of a generated design, or fetch account history), but these fall outside the intended handoff where the user completes the workflow externally.