UK Visa Sponsor MCP Server

What it does#

This is the AI-native interface to the UK Register of Licensed Sponsors. It wraps the same 125,000-company dataset that powers UK Visa Sponsor Search and exposes it to any Model Context Protocol client (Claude Desktop, Cursor, or a remote agent), so sponsorship questions can be answered in natural language.

Tools#

Four tools are implemented in server.py on top of FastMCP:

The tools are built to compose: an assistant can search, narrow by route or location, then pull details or stats without the user spelling out each step.

Transport modes#

The server runs over either MCP transport:

# Local, stdio (Claude Desktop / Cursor)
fastmcp run server.py
 
# Remote, Server-Sent Events
fastmcp run server.py --transport sse --host 0.0.0.0 --port 8000

Remote clients connect at http://your-host:8000/sse.

Configuration#

The data source is read from an environment variable, defaulting to the public API:

Registering it with Claude Desktop:

{
  "mcpServers": {
    "uk-visa-sponsors": {
      "command": "fastmcp",
      "args": ["run", "/absolute/path/to/server.py"]
    }
  }
}

Stack and deployment#

Python 3.11+ on FastMCP, packaged with pyproject.toml and installable via pip install -e .. A Dockerfile is included for containerised SSE deployments (docker build -t uk-visa-sponsor-mcp .), and fastmcp dev server.py launches the FastMCP Inspector for interactive tool testing. Licensed MIT.

Why it matters#

The engineering value is in tool design, not API plumbing: clear tool names, typed parameters, and descriptions that tell the model when and why to use each call. That is what turns a static dataset into something an agent can query reliably.