Let your AI agent try clothes on for you.
One MCP endpoint and Claude, Cursor, or any MCP client can call our virtual try-on engine directly — send a photo of a person and a garment, and the result comes back right in the conversation.
{
"mcpServers": {
"tryonanytime": {
"type": "http",
"url": "https://tryonanytime.com/api/mcp",
"headers": { "Authorization": "Bearer sk_your_key" }
}
}
}What is MCP?
Model Context Protocol (MCP) is an open standard that lets AI assistants call external tools. TryOnAnytime exposes its virtual try-on engine as an MCP server, so your agent can dress your photos without you ever leaving the chat.
Same engine as the REST API
MCP is a protocol layer on top of the same try-on pipeline — identical quality, identical credits. Use REST when you script; use MCP when your agent does the talking.
Connect in one minute
Point your MCP client at a single HTTP endpoint and authenticate with your API key.
Claude Desktop / Claude Code
Add the server with one CLI command — or paste the JSON into claude_desktop_config.json.
Cursor
Drop this into .cursor/mcp.json (project) or ~/.cursor/mcp.json (global), then enable the server in Settings → MCP.
Any Streamable HTTP client
The server speaks MCP over Streamable HTTP. Point any compliant client at the URL with the Bearer header.
claude mcp add --transport http tryonanytime \ https://tryonanytime.com/api/mcp \ --header "Authorization: Bearer sk_your_key"
Legacy clients (mcp-remote)
For clients that only support stdio, bridge with mcp-remote:
npx mcp-remote https://tryonanytime.com/api/mcp \ --header "Authorization: Bearer sk_your_key"
Tools reference
Three tools. That's the whole surface.
virtual_tryonSubmit a virtual try-on. When the render finishes within the call, the result image is returned directly as MCP image content.
virtual_tryon({
"person_url": "https://…/me.jpg", // required
"garment_url": "https://…/jacket.jpg", // required
"prompt": "keep my face and pose" // optional
})get_tryon_resultPoll a task's status and fetch the result image once it succeeds.
get_tryon_result({
"task_id": "a1b2c3d4-e5f6-4a7b-8c9d-0e1f2a3b4c5d"
})get_credit_balanceCheck the credit balance on the current API key.
get_credit_balance() // no arguments
See it in a conversation
No new UI, no tab switching — just ask.
Done — here's you in the jacket. This try-on used 10 credits; you have 32 left.
Billing & limits
10 credits per try-on
Every virtual_tryon call deducts 10 credits from your account — same price as the REST API. Top up or subscribe any time.
Rate limit
virtual_tryon: one call every 5 seconds per key owner (the account the key belongs to). Reads (get_tryon_result, get_credit_balance): 500 ms per key owner, 1 second per IP. Agents should poll get_tryon_result instead of spamming virtual_tryon.
Key management
Create and revoke keys in your dashboard. One key works across MCP, the REST API, and the browser extension.
Keep building
Give your agent try-on superpowers
One MCP endpoint, three tools, the same credits as the REST API. Connect your client and let your agent dress the photos.