OfferQuest MCP Server (mcp_j.py)

A fast, secure, and LLM-friendly Model Context Protocol (MCP) server that scrapes job listings from major platforms (LinkedIn, Indeed, Google) and converts them into structured Markdown format.

Features

• Blazing Fast API Scraping: Uses python-jobspy to pull latest jobs instantly without heavy browser automation overhead.

• Multi-Search Support: Automatically handles parallel searching for multiple comma-separated job titles and locations in a single unified run.

• LLM-Optimized Output: Jobs are formatted into a clean, easy-to-read Markdown table specifically designed for AI agents and LLMs to parse and understand securely.

• Strict Security:

  • All user inputs are sanitized to drop executable scripts and strange characters (_sanitize_text).

  • Limits max input length to prevent denial-of-service (DoS).

  • Internal errors/stack traces are masked from the user to prevent data leakage.

• Granular Targeting:

  • Dynamic Country selection explicitly prevents the APIs from serving out-of-bounds global results.

  • "Max Hours Old" filter perfectly isolates ultra-fresh job postings.

Related MCP server: LinkedIn MCP Server

Installation

Ensure you have Python 3.10+ installed.

1. Clone or navigate to this project directory.

2. Create a virtual environment (Recommended):

   python3 -m venv myenv
   source myenv/bin/activate

3. Install dependencies:

   pip install -r requirements.txt

Usage

Start the server locally:

python3 mcp_j.py

• The server will boot up a local Gradio interface (usually http://127.0.0.1:7860).

• If you are plugging this into an MCP client, the endpoint is exposed at /gradio_api/mcp/.

Deployment

Since the codebase is stateless and doesn't rely on background Playwright Chromium browsers, this script is highly viable for lightweight containerized deployments (Docker, Render, Heroku) or standard VPS setups.

Security checklist completed for deployment:

• Catch-all error blocks to hide raw API tracebacks.

• Built-in input sanitization using rigorous Regex.

• Hard limits on payload size (max_results=25 upper-bound).

Note: For highest stability on cloud providers, ensure that the IP address you are querying from isn't strictly blacklisted by Indeed/LinkedIn.