Virtual try-on, inside your AI assistant.

Add Genlook to any MCP-compatible assistant and try clothes on in plain language. The server lives at:

https://mcp.genlook.app

No API key needed. The connector authenticates with OAuth. Sign in (or create an account) when your assistant prompts you, and you're ready to generate.

5

Free credits to start, no card required.

1 credit

Per try-on generation.

OAuth

No API keys to copy, paste, or rotate.

~10s

From prompt to on-body result in the chat.

01 · Connect

Three steps, no keys.

1

Add the connector

Point your assistant at https://mcp.genlook.app as a custom MCP / HTTP connector.

2

Sign in with OAuth

Approve access with your Genlook account, no keys to manage or rotate.

3

Start generating

Ask your assistant to try a garment on a photo, and the result appears right in the chat.

Set it up in your client

Claude

1. Open claude.ai → Settings → Connectors
2. Click Add custom connector and paste https://mcp.genlook.app
3. Sign in with your Genlook account when prompted

Claude Code

claude mcp add --transport http genlook https://mcp.genlook.app

Then run /mcp inside Claude Code to complete the OAuth sign-in.

ChatGPT

1. Settings → Apps & Connectors → enable Developer mode
2. Create a connector with https://mcp.genlook.app (OAuth)
3. Sign in with your Genlook account when prompted

Cursor

Add to ~/.cursor/mcp.json (or via Settings → MCP → Add new MCP server):

{
  "mcpServers": {
    "genlook": {
      "url": "https://mcp.genlook.app"
    }
  }
}

Cursor handles the OAuth flow on first use.

Tools

The tools mirror the REST API:

Tool	What it does
upload_photo	Upload a person photo (public URL or base64). Returns an imageId to reuse for try-ons.
get_uploaded_photo	Return your most recent non-expired upload so you don't have to re-upload between conversations.
generate_try_on	Generate a try-on from a garment image URL on a given person photo. Waits and returns the result.
get_generation	Poll a generation by id; returns the image once it's complete.
get_account	Show remaining credits and the top-up URL.

All image URLs passed to the tools must be publicly accessible, no login-gated or private links.

Example prompts

• "Here's a dress I like: <image URL>, show me wearing it."
• "Use this photo of me from now on: <image URL>."
• "Try these sneakers on me, then compare with the previous pair."
• "How many try-on credits do I have left?"

The same engine that powers the Genlook storefront widget runs behind the MCP tools. A single photo plus a garment URL is enough to get a clean, on-body result back in the conversation. Uploaded photos and generated images follow the account retention window (7 days by default), then are deleted automatically.

Building for a storefront instead?

Try-On API for developers →Shopify app →WooCommerce plugin →

Same engine, same account

02 · MCP, specifically

Asked before connecting.

Do I need an API key?↓

No. The MCP connector uses OAuth. Sign in with your Genlook account and you're authenticated. There are no keys to copy or rotate.

How much does it cost?↓

New accounts start with 5 free credits, and each generation costs 1 credit. Top up anytime at platform.genlook.app/topup.

Which clients work?↓

Any MCP-compatible client over Streamable HTTP, including Claude, Claude Code, ChatGPT (developer mode), and Cursor.

Is it the same engine as the storefront widget?↓

Yes. The MCP tools call the same Virtual Try-On API and the same Account and credits as platform.genlook.app.

Try it on from your AI assistant.

Connect the MCP server and start with 5 free credits, top up anytime.

Get free creditsExplore the API