Adzuna Job Search MCP Server

What it does#

A Model Context Protocol server that gives AI assistants live, structured access to the Adzuna job-search API across 12 countries. Published to PyPI as adzuna-mcp, it turns labour-market data (listings, salaries, employers) into typed tools an assistant can chain.

Tools#

Seven tools are exposed over stdio transport:

(* required.) The tool descriptions steer the model to call get_categories first so search_jobs receives a valid category tag, and they encode real constraints: salaries are annual and in local currency, and many listings have no salary data, so salary filters drop unlisted roles.

Supported markets#

Twelve countries with local-currency salaries: United Kingdom, United States, Germany, France, Australia, New Zealand, Canada, India, Poland, Brazil, Austria, and South Africa.

Install and configure#

pip install adzuna-mcp      # or: uvx adzuna-mcp

Credentials come from the free Adzuna developer tier via environment variables (ADZUNA_APP_ID, ADZUNA_APP_KEY). A Claude Desktop entry:

{
  "mcpServers": {
    "adzuna-jobs": {
      "command": "/path/to/venv/bin/python",
      "args": ["/path/to/server.py"],
      "env": { "ADZUNA_APP_ID": "your_app_id", "ADZUNA_APP_KEY": "your_app_key" }
    }
  }
}

Engineering#

The project ships a test suite under tests/ and two GitHub Actions workflows: ci.yml for continuous integration and publish.yml for automated PyPI releases. A Dockerfile supports containerised runs, and fastmcp dev server.py opens an interactive inspector for trying the tools. Licensed MIT.