Furniture search.
Inside your assistant.
Room×Lab Furniture Search helps your AI assistant discover furniture across retailers, compare prices and dimensions, and explore similar pieces. Every result links back to Room×Lab so you can take a closer look.
Already use an AI assistant? Bring Room×Lab along. This optional connection is for people who prefer searching in a conversation. Prefer to browse? Explore the shop as usual—no setup needed.
Every option connects to the same read-only furniture search. No Room×Lab account or API key is needed. Choose one connection method per assistant.
Connect to Claude
- In Claude, open Customize → Connectors and choose Add custom connector.
- Name it Room×Lab Furniture Search and enter this server URL:
https://roomxlab.com/mcp. - Leave authentication unset. Add the connector, then enable it in your conversation’s connectors menu.
No API key or Room×Lab account is needed. This connection only reads public furniture information; it cannot place orders or change your accounts. For managed Claude workspaces, ask your workspace owner to add it.
Claude Code setup
claude mcp add --transport http roomxlab https://roomxlab.com/mcp
Our Claude package on GitHub also includes plugin installation instructions.
Connect to Codex
Add Room×Lab as a remote MCP server in your local Codex configuration. MCP is the connection standard that lets your assistant search Room×Lab.
- Open
~/.codex/config.tomland add the configuration below, keeping your existing settings. - Restart your Codex session to load the connection. No authentication token is needed.
- Ask Codex to use Room×Lab to find or compare furniture.
[mcp_servers.roomxlab]
url = "https://roomxlab.com/mcp"
Codex’s current setup guide · Codex package on GitHub
Codex CLI shortcut
Instead of editing the configuration by hand, run:
codex mcp add roomxlab --url https://roomxlab.com/mcp
More assistants, same furniture search
Use these client-specific configurations for other popular agents. Merge the Room×Lab entry into existing settings rather than replacing other servers.
Cursor
Add this to .cursor/mcp.json in your project, or ~/.cursor/mcp.json for all projects. Enable the server in Cursor’s MCP settings.
{
"mcpServers": {
"roomxlab": {
"url": "https://roomxlab.com/mcp"
}
}
}
VS Code / GitHub Copilot
For VS Code’s built-in chat, add this to your workspace’s .vscode/mcp.json. Start the server from the configuration editor, review the trust prompt, and enable its tools in chat.
{
"servers": {
"roomxlab": {
"type": "http",
"url": "https://roomxlab.com/mcp"
}
}
}
Gemini CLI
Add this to ~/.gemini/settings.json, then restart Gemini CLI. Keep normal tool confirmation enabled.
{
"mcpServers": {
"roomxlab": {
"httpUrl": "https://roomxlab.com/mcp"
}
}
}
Another MCP-compatible assistant
Add a remote MCP server with Streamable HTTP, URL https://roomxlab.com/mcp, and no authentication. Configuration field names vary by client; follow your assistant’s documentation.
These instructions follow the clients’ documented connection formats; we have not tested every client version. Availability depends on your assistant’s plan and administrator settings. This endpoint is not a webpage, and a manual connection does not imply listing in an assistant’s public directory.
Try asking
- Find a walnut coffee table under $1,500.
- Show me Article sofas, least expensive first.
- Compare the prices and dimensions of these two Room×Lab pieces.
- Find similar furniture from another retailer.
Prices and product details can change. A missing price is not proof that a piece meets your budget. Stock information is intentionally omitted; confirm price, availability, delivery, and fit with the retailer before purchasing.
What you can do
Search across brands, look up product details, compare pieces, explore similar designs, and browse a particular retailer.
Technical tool reference
search_furniture- Search by furniture category, retailer, budget, dimensions, colors, styles, and materials. Returns up to eight pieces.
get_furniture_product- Read concise facts for a Room×Lab product ID returned by another tool.
compare_furniture_products- Compare two to six known product IDs, including captured prices and dimensions.
find_similar_furniture- Explore similar pieces and, where available, same-item listings. Similarity is not a guarantee of identical construction or quality.
browse_furniture_retailer- Search a named retailer's catalog with optional furniture filters.
The connector cannot purchase furniture, contact retailers, save items, or change accounts. It returns public product facts, images, and Room×Lab links, not private catalog evidence or a downloadable catalog.
Advanced: agent query guide
Use the exact field names and allowed values advertised by tools/list. Put category, retailer, budget, dimensions, color, style, and material constraints in their structured fields. Use query only for a short remaining literal term, such as a model name or wood species. For “walnut coffee table under $1,500,” call search_furniture with:
{
"query": "walnut",
"category": "coffee-table",
"materials": ["wood"],
"maxPrice": 1500
}
A category normally makes a separate room filter unnecessary. Dimensions use inches; multiple color, style, or material values mean any selected value. Omit sorting for relevance. Use returned product IDs for follow-up calls, and preserve any notes explaining relaxed matches or uncertain prices.
Limits and troubleshooting
The connector is for individual shopping requests, not scraping or bulk extraction. Searches require a query, category, room, or retailer and do not offer pagination or full catalog totals. Please follow our terms of use.
If a request returns HTTP 429, wait for the duration in Retry-After before retrying. Avoid parallel retry loops. For a validation error, use the published schema; for no matches, narrow or clarify the shopper's request rather than repeatedly submitting broad searches.
Do not put personal or confidential information in search inputs. See our privacy policy for request processing and abuse-prevention details. For help, email contact@roomxlab.com with the tool name and error, without credentials or private conversation content.