bhishman-desai/resume-mcp

Resume MCP Server

A Model Context Protocol (MCP) server that hosts your resume data in PostgreSQL. This server provides tools to manage your resume with version control, validation, and safety features.

Features

• Resume Management: Store and manage your resume in PostgreSQL
• Version Control: Automatic snapshots on every update with restore capability
• Validation: All updates validated using Zod schemas
• Safety Features: Automatic backups, path traversal protection, and error handling
• SOLID Architecture: Clean separation of concerns with modular structure
• HTTPS Ready: Deployable on Render with HTTPS support via SSE transport

Tools

get_resume

Returns the current resume JSON. No authentication required.

update_resume

Replace the entire resume. Requires { resume, apiKey }.

patch_resume

Merge partial resume data with existing resume. Requires { partialResume, apiKey }.

list_versions

List all snapshot filenames. No authentication required.

restore_version

Restore resume from a snapshot. Requires { filename, apiKey }.

Installation

1. Clone the repository:

git clone https://github.com/bhishman-desai/resume-mcp.git
cd resume-mcp

2. Install dependencies:

npm install

3. Set up environment variables:

cp .env.example .env
# Edit .env with your configuration

Configuration

Environment Variables

• DATABASE_URL: PostgreSQL connection string (preferred for cloud providers)
• DB_HOST, DB_PORT, DB_NAME, DB_USER, DB_PASSWORD: Individual database connection parameters
• DB_SSL: Set to true for SSL connections
• API_KEY: API key for protecting update/restore operations
• PORT: Server port (default: 3000)
• HOST: Server host (default: 0.0.0.0)

Local Development

Running as MCP Server (stdio mode)

For local development with Cursor, run:

npm start

The server will automatically detect stdio mode and connect via standard input/output.

Running as HTTP Server

To test HTTP mode locally:

NODE_ENV=production npm start

The server will start on http://localhost:3000 with SSE endpoint at /sse.

Deployment on Render

1. Database Setup

1. Create a PostgreSQL database on Render
2. Copy the Internal Database URL from Render dashboard

2. Web Service Setup

1. Create a new Web Service on Render
2. Connect your GitHub repository
3. Configure the service:
   • Build Command: npm install
   • Start Command: npm start
   • Environment: Node

3. Environment Variables

Add these environment variables in Render dashboard:

DATABASE_URL=<your_postgres_internal_url>
API_KEY=<your_secure_api_key>
PORT=10000

Render automatically provides HTTPS, so your MCP server will be accessible via HTTPS.

4. Connecting to Cursor

The MCP server supports two connection modes:

Local Development (stdio)

For local development, Cursor automatically connects via stdio when you run:

npm start

Remote Deployment (SSE)

For connecting to your deployed server on Render:

1. Get your Render service URL (e.g., https://your-service.onrender.com)
2. Configure Cursor to connect to the remote server. The exact configuration depends on your Cursor version, but typically involves:
   • Adding the server URL in Cursor's MCP settings
   • The server exposes an SSE endpoint at /sse for MCP connections

Note: The MCP server uses SSE transport which allows Cursor to connect without requiring an API key for the connection itself. However, write operations (update_resume, patch_resume, restore_version) still require the API key as a parameter in the tool call.

Project Structure

resume-mcp/
├── src/
│   ├── config/          # Configuration management
│   ├── database/         # PostgreSQL connection and operations
│   ├── validation/       # Zod schemas for validation
│   ├── tools/            # MCP tool implementations
│   └── utils/            # Utility functions (security, error handling)
├── app.js                # Main server entry point
├── package.json
└── .env.example

Safety Features

Validation

• All resume updates are validated using Zod schemas
• PATCH operations validate partial inputs and merged results
• Invalid data triggers automatic backups

Automatic Backups

• Every update creates a timestamped backup
• Emergency backups created on validation failures
• Pre-restore backups created before version restoration

Security

• API key protection for write operations
• Filename sanitization to prevent path traversal
• Input validation on all operations

Error Handling

• Graceful error handling with user-friendly messages
• Server continues running even with malformed data
• Automatic recovery mechanisms

Resume Schema

The resume follows a flexible schema that includes:

• Personal information (name, email, phone, location)
• Summary
• Experience (array of work experiences)
• Education (array of education entries)
• Skills (array of strings)
• Certifications (array of certification objects)
• Projects (array of project objects)
• Languages (array of language objects)

The schema uses .passthrough() to allow additional custom fields.

License

MIT

Author

Bhishman Desai