MCP-LinkedIn

Instructions

Implementation Reference

• linkedin_mcp_server/tools/job.py:81-105 (handler)

  The core handler function for the 'get_recommended_jobs' tool. It uses LinkedIn's JobSearch class configured to scrape recommended jobs and returns a list of job dictionaries. Includes error handling.

  @mcp.tool()
  async def get_recommended_jobs() -> List[Dict[str, Any]]:
      """
      Get your personalized recommended jobs from LinkedIn
  
      Returns:
          List[Dict[str, Any]]: List of recommended jobs
      """
      try:
          driver = safe_get_driver()
  
          logger.info("Getting recommended jobs")
          job_search = JobSearch(
              driver=driver,
              close_on_complete=False,
              scrape=True,  # Enable scraping to get recommended jobs
              scrape_recommended_jobs=True,
          )
  
          if hasattr(job_search, "recommended_jobs") and job_search.recommended_jobs:
              return [job.to_dict() for job in job_search.recommended_jobs]
          else:
              return []
      except Exception as e:
          return handle_tool_error_list(e, "get_recommended_jobs")

• linkedin_mcp_server/server.py:26-29 (registration)

  Registration point where register_job_tools(mcp) is called, which in turn registers the get_recommended_jobs tool via @mcp.tool() decorator.

  # Register all tools
  register_person_tools(mcp)
  register_company_tools(mcp)
  register_job_tools(mcp)