MCP Server - U2L AI Developer Hub

Connect AI assistants like Claude, Cursor, and ChatGPT to U2L. Create, search, and manage your short links from any MCP-capable client.

U2L hosts a remote MCP (Model Context Protocol) server, so AI assistants can manage your links for you — “shorten this URL”, “how many clicks did my launch link get”, “clean up my old campaign links”.

Endpoint — https://u2l.ai/api/mcp (Streamable HTTP)

Authentication uses your regular API key, sent as an Authorization: Bearer header. Create one at Settings → API — available on every plan.

Setup

claude mcp add --transport http u2l https://u2l.ai/api/mcp \
  --header "Authorization: Bearer u2l_live_your_api_key"

{
  "mcpServers": {
    "u2l": {
      "url": "https://u2l.ai/api/mcp",
      "headers": {
        "Authorization": "Bearer u2l_live_your_api_key"
      }
    }
  }
}

{
  "servers": {
    "u2l": {
      "type": "http",
      "url": "https://u2l.ai/api/mcp",
      "headers": {
        "Authorization": "Bearer u2l_live_your_api_key"
      }
    }
  }
}

Available tools

Tool	What it does
`create_link`	Create a short link or QR code (custom alias, domain, title)
`get_link`	Fetch one link by domain + slug
`list_links`	List/search your links with filters and pagination
`update_link`	Change a link’s destination, title, or tags
`delete_link`	Delete a link permanently
`get_link_analytics`	Click totals and breakdowns for a link (Advanced+)
`list_domains`	List your available short domains
`get_account_stats`	Account totals, plan, and limits

Every tool call runs through the same /api/v1 endpoints as the REST API — identical auth, plan gates, and rate limits. Nothing is possible over MCP that isn’t possible with your API key directly.

Notes

The server is stateless: no sessions, no OAuth dance — just the header. If your client asks about OAuth, choose header/API-key auth instead.
A missing or invalid key doesn’t break the connection; tool calls return an instruction to configure the Authorization header.
Deleting links via an AI assistant is permanent. Assistants ask before destructive actions, but scope your expectations (and revoke the key anytime from Settings → API).