jobspy

JobSpy MCP Server exposes an MCP interface for searching jobs across multiple platforms using the JobSpy tool. It supports real-time updates over Server-Sent Events (SSE) and can accept tool invocation requests via the MCP protocol, returning structured job data that can be consumed by AI assistants or web clients. The server is designed to integrate with Claude or similar agents via a Claude-friendly config and also provides standard HTTP endpoints for web clients. Core capabilities include searching across multiple sites (Indeed, LinkedIn, Glassdoor, etc.), filtering results by terms, location, time frames, and outputting results in JSON or CSV formats. Tools exposed by the MCP server include a search_jobs tool that aggregates results from various job sites and returns a unified response suitable for downstream processing.

To use it, run the MCP server (see installation steps) and connect clients via the provided endpoints. For Claude integration, configure the MCP server in Claude’s desktop config with the Node.js runtime and the path to the server entry point. Web clients can connect to the SSE transport at the /mcp/connect endpoint and submit requests to /mcp/request, receiving tool responses and progress updates in real time.

Prerequisites:

• Node.js 16+ (for the MCP server runtime)
• Python 3.6+ (for the JobSpy tool and any Python-based setup)
• The JobSpy tool installed and accessible in the environment

Installation steps:

1. Clone the MCP server repository and install dependencies

# Clone the repository
git clone https://github.com/yourusername/jobspy-mcp-server.git
cd jobspy-mcp-server

# Install Node.js dependencies
npm install

2. Install and configure the JobSpy tool

# Ensure the JobSpy tool is available
# Depending on the project layout, install Python dependencies for JobSpy
cd ../jobSpy
pip install -r requirements.txt
chmod +x run.sh

3. Run the MCP server

# From the server root
npm start

4. Optional: run in Docker (if Dockerfile provided)

# Build the Docker image
docker build -t jobspy-mcp-server .

# Run the container
docker run -p 9423:9423 jobspy-mcp-server

Environment variables commonly used:

• JOBSPY_DOCKER_IMAGE: Docker image to use for JobSpy (default: jobspy)
• JOBSPY_ACCESS_TOKEN: Access token for JobSpy API (optional)
• PORT: MCP server port (default: 9423)
• HOST: MCP server host (default: 0.0.0.0)
• ENABLE_SSE: Enable SSE transport (1 to enable, 0 to disable; default 0)

Tips and common issues:

• Ensure the JobSpy script is discoverable by the server (look for ../jobSpy/run.sh, ./run.sh, or /app/run.sh in Docker environments).
• If SSE transport isn’t working, verify ENABLE_SSE is set to 1 and that the server binding matches the client URL.
• When configuring Claude, point to the server’s Node entry and ensure node path matches your deployment environment.
• The server can output results in JSON or CSV; verify the client expects the correct format via the tool parameters.