linkedin-job-search-server

Description

This is an MCP server that provides a tool for searching jobs on LinkedIn using natural language. The tool, chat_linkedin_jobs, allows users to input a natural language query, which is then used to scrape job listings from LinkedIn.

Requirements

• The tool should provide a chatbot-like conversational experience for job searching.
• The tool should intelligently search LinkedIn based on natural language input, extracting keywords, location, and other relevant criteria.
• The tool should be able to ask clarifying questions if the user's input is ambiguous.
• The tool should present a list of relevant job postings from LinkedIn.
• The tool should be implemented as an MCP server.

Approach and Architecture

The linkedin-job-search-server is implemented as a Node.js application using the @modelcontextprotocol/sdk to create an MCP server. The server exposes a single tool, chat_linkedin_jobs, which handles the job search functionality.

The current implementation uses web scraping to fetch job listings from LinkedIn. It utilizes the axios library to make HTTP requests and the cheerio library to parse the HTML content of LinkedIn job search pages.

The tool currently has a basic caching mechanism to reduce the number of requests made to LinkedIn and avoid rate limiting.

Dependencies

• @modelcontextprotocol/sdk: The Model Context Protocol SDK for creating MCP servers.
• axios: For making HTTP requests to LinkedIn.
• cheerio: For parsing HTML content and extracting job listing information.
• compromise: For basic natural language processing (currently not fully utilized).
• random-useragent: To generate random user agents for web scraping.
• typescript: For type checking and better code maintainability.

Usage

To use the chat_linkedin_jobs tool, you can send a request to the MCP server with the following format:

<use_mcp_tool>
<server_name>linkedin-job-search-server</server_name>
<tool_name>chat_linkedin_jobs</tool_name>
<arguments>
{
  "query": "your job search query here"
}
</arguments>
</use_mcp_tool>

Replace "your job search query here" with your desired job search query, such as "software engineer jobs in London".

Limitations

• The current implementation uses a placeholder for natural language processing and simply returns the user's query.
• The web scraping approach is brittle and may break if LinkedIn changes its website structure.
• There is no authentication implemented, so the tool can only access publicly available job listings.
• The tool may encounter rate limits from LinkedIn if too many requests are made in a short period.
• There is no integration with the user's CV or LinkedIn profile to personalize search results.

Future Enhancements

• Implement more sophisticated natural language processing using compromise or other NLP libraries to better understand user queries and extract relevant search parameters.
• Improve the web scraping logic to handle various search filters and extract more detailed job information.
• Implement proper authentication with the LinkedIn API (if possible) to access more features and potentially avoid rate limiting.
• Add more robust error handling and rate limit management.
• Integrate with the user's CV and LinkedIn profile to personalize search results.
• Add the ability to assist with job applications.

Build and Run

To build and run the server, use the following commands:

npm install
npm run build

This will install the dependencies, compile the TypeScript code, and make the server executable. The server will then be accessible through the MCP client.