What is the snapy MCP server?

It is a Model Context Protocol (MCP) server that adds a publish tool to your AI client. The assistant can take a page or file it made and host it on snapy, then return a shareable link.

An HTML page, a file, or a static site (HTML, CSS, JavaScript). You get back a live link plus a private analytics link.

Do I need an MCP client?

No. MCP clients like Claude Desktop and Cursor can use the MCP server, but any AI agent or script can call the snapy publish API directly. Both are covered below.

Yes. Publishing a page or file and getting a link is free, with no signup.

MCP server

snapy MCP server

Give your AI assistant the power to publish. With snapy, Claude, Cursor, or any AI agent can host an HTML page, a file, or a static site and hand back a live link, automatically.

What the snapy MCP server does

snapy turns a file or page into a link. The MCP server brings that into your AI assistant. After you connect it, the assistant gets a publish tool. When it builds something, an HTML page, a chart, a small static site, it can host it on snapy and return a live URL, with no copy-paste and no manual upload.

This is built on the Model Context Protocol (MCP), the open standard for giving an AI assistant new tools. So it works with any MCP client, including Claude Desktop and Cursor.

Add the snapy MCP server to your AI client

In your MCP client config (for example Claude Desktop's config file), add a snapy entry:

{
  "mcpServers": {
    "snapy": {
      "command": "npx",
      "args": ["-y", "snapy-mcp"]
    }
  }
}

The server is open source, so you can also run it straight from the repository instead of npx: set command to node and args to the path of the server's index.js. Either way, restart your client and the publish tool appears.

The publish tool

Once connected, the assistant can call publish:

content (required): the text or HTML to host.
name (optional): a custom link name, like launch for launch.snapy.page.
filename (optional): defaults to index.html.

It returns the live link and a private analytics link. So you can say "build a landing page and publish it," and the assistant hosts it and gives you the URL.

No MCP client? Use the publish API (any AI agent)

Any AI tool, agent, or script can publish to snapy over plain HTTP, no MCP required. This is the simplest way to let an automation or a custom GPT host files and pages.

POST https://api.snapy.host/api/publish
Content-Type: application/json

{ "content": "<h1>Hello</h1>", "name": "my-page", "filename": "index.html" }

The response includes the live url and a private stats_url. There is a machine-readable OpenAPI schema at https://api.snapy.host/openapi.json and https://snapy.host/openapi.json, so you can wire it into a ChatGPT custom GPT action or any agent that reads OpenAPI.

What you can publish

An HTML page or landing page the AI generated.
A file you want shared as a link.
A static site (HTML, CSS, JavaScript) zipped together.
A chart, a report, a one-pager, or any artifact your assistant produces.

Why this is useful

The hard part of AI-generated work is sharing it. A chat link often needs the other person to log in, and the output is stuck inside the tool. snapy gives it a clean public link that opens for anyone, on any device, instantly.

New to snapy? See how snapy works with Claude and ChatGPT, the developer and AI guide, or just host an HTML file free.