Candidate Server

A Model Context Protocol (MCP) server that gives LLMs access to information about a candidate.

Overview

Important: This server is intended to be used as a library to be integrated into other applications, not as a standalone service. The provided startup methods are for demonstration and testing purposes only.

Resources

This MCP server provides the following resources:

• candidate-info://resume-text: Resume content as text
• candidate-info://resume-url: URL to the resume
• candidate-info://linkedin-url: LinkedIn profile URL
• candidate-info://github-url: GitHub profile URL
• candidate-info://website-url: Personal website URL
• candidate-info://website-text: Content from the personal website

Tools

This MCP server also provides tools that return the same candidate information:

• get_resume_text: Returns the candidate's resume content as text
• get_resume_url: Returns the URL to the candidate's resume
• get_linkedin_url: Returns the candidate's LinkedIn profile URL
• get_github_url: Returns the candidate's GitHub profile URL
• get_website_url: Returns the candidate's personal website URL
• get_website_text: Returns the content from the candidate's personal website
• contact_candidate: Sends an email to the candidate (requires Mailgun configuration)

Usage

npm install @jhgaylor/candidate-mcp-server

Library Usage

This package is designed to be imported and used within your own applications.

Stdio

Starting the process is a breeze with stdio. The interesting part is providing the candidate configuration.

Where you source the candidate configuration is entirely up to you. Maybe you hard code it. Maybe you take a JSONResume url when you start the process. It's up to you!

import { createServer } from '@jhgaylor/candidate-mcp-server';
import { StdioServerTransport } from "@modelcontextprotocol/sdk/server/stdio.js";

// Configure your server
const serverConfig = { 
  name: "MyCandidateServer", 
  version: "1.0.0",
  mailgunApiKey: process.env.MAILGUN_API_KEY,
  mailgunDomain: process.env.MAILGUN_DOMAIN
};
const candidateConfig = { 
  name: "John Doe",
  email: "[email protected]", // Required for the contact_candidate tool
  resumeUrl: "https://example.com/resume.pdf",
  // other candidate properties
};

// Create server instance
const server = createServer(serverConfig, candidateConfig);

// Connect with your preferred transport
await server.connect(new StdioServerTransport());
// or integrate with your existing HTTP server

StreamableHttp

Using the example code provided by the typescript sdk we can bind this mcp server to an express server.

import express from 'express';
import { Request, Response } from 'express';
import { createServer } from '@jhgaylor/candidate-mcp-server';
import { StreamableHTTPServerTransport } from "@modelcontextprotocol/sdk/server/streamablehttp.js";

// Configure your server
const serverConfig = { 
  name: "MyCandidateServer", 
  version: "1.0.0",
  mailgunApiKey: process.env.MAILGUN_API_KEY,
  mailgunDomain: process.env.MAILGUN_DOMAIN,
  contactEmail: "[email protected]",
};
const candidateConfig = { 
  name: "John Doe",
  resumeUrl: "https://example.com/resume.pdf",
  // other candidate properties
};

// Factory function to create a new server instance for each request
const getServer = () => createServer(serverConfig, candidateConfig);

const app = express();
app.use(express.json());

app.post('/mcp', async (req: Request, res: Response) => {
  // In stateless mode, create a new instance of transport and server for each request
  // to ensure complete isolation. A single instance would cause request ID collisions
  // when multiple clients connect concurrently.
  
  try {
    const server = getServer(); 
    const transport = new StreamableHTTPServerTransport({
      sessionIdGenerator: undefined,
    });
    res.on('close', () => {
      console.log('Request closed');
      transport.close();
      server.close();
    });
    await server.connect(transport);
    await transport.handleRequest(req, res, req.body);
  } catch (error) {
    console.error('Error handling MCP request:', error);
    if (!res.headersSent) {
      res.status(500).json({
        jsonrpc: '2.0',
        error: {
          code: -32603,
          message: 'Internal server error',
        },
        id: null,
      });
    }
  }
});

app.get('/mcp', async (req: Request, res: Response) => {
  console.log('Received GET MCP request');
  res.writeHead(405).end(JSON.stringify({
    jsonrpc: "2.0",
    error: {
      code: -32000,
      message: "Method not allowed."
    },
    id: null
  }));
});

app.delete('/mcp', async (req: Request, res: Response) => {
  console.log('Received DELETE MCP request');
  res.writeHead(405).end(JSON.stringify({
    jsonrpc: "2.0",
    error: {
      code: -32000,
      message: "Method not allowed."
    },
    id: null
  }));
});

// Start the server
const PORT = process.env.PORT || 3000;
app.listen(PORT, () => {
  console.log(`MCP Stateless Streamable HTTP Server listening on port ${PORT}`);
});

Express

Instead of writing the binding between express and the mcp transport yourself, you can use express-mcp-handler to do it for you.

npm install express-mcp-handler

import express from 'express';
import { statelessHandler } from 'express-mcp-handler';
import { createServer } from './server';

// You can configure the server factory to include Mailgun settings
const createServerWithConfig = () => {
  const serverConfig = { 
    name: "MyCandidateServer", 
    version: "1.0.0",
    mailgunApiKey: process.env.MAILGUN_API_KEY,
    mailgunDomain: process.env.MAILGUN_DOMAIN,
    contactEmail: "[email protected]",
  };
  const candidateConfig = { 
    name: "John Doe",
    resumeUrl: "https://example.com/resume.pdf",
    // other candidate properties
  };
  
  return createServer(serverConfig, candidateConfig);
};

// Configure the stateless handler
const handler = statelessHandler(createServerWithConfig);

// Create Express app
const app = express();
app.use(express.json());

// Mount the handler (stateless only needs POST)
app.post('/mcp', handler);

// Start the server
const PORT = process.env.PORT || 3002;
app.listen(PORT, () => {
  console.log(`Stateless MCP server running on port ${PORT}`);
});

Development

# Install dependencies
npm install

# Build the project
npm run build

# Run in development mode with auto-restart
npm run dev

Demo / Debug Startup via stdio

# Start with STDIO (demo only)
npm start

When running with STDIO, you can interact with the server by sending MCP messages as single-line JSON objects:

# Example of sending an initialize message via STDIO
echo '{"jsonrpc": "2.0","id": 1,"method": "initialize","params": {"protocolVersion": "2024-11-05","capabilities": {"roots": {"listChanged": true},"sampling": {}},"clientInfo": {"name": "ExampleClient","version": "1.0.0"}}}' | node dist/index.js --stdio

# List resources
echo '{"jsonrpc": "2.0","id": 2,"method": "resources/list","params": {}}' | node dist/index.js --stdio

# Access a resource
echo '{"jsonrpc": "2.0","id": 3,"method": "resources/read","params": {"uri": "candidate-info://resume-text"}}' | node dist/index.js --stdio

# List Tools
echo '{"jsonrpc": "2.0","id": 2,"method": "tools/list","params": {}}' | node dist/index.js --stdio

# Call a tool
echo '{"jsonrpc": "2.0","id": 4,"method": "tools/call","params": {"name": "get_resume_text", "args": {}}}' | node dist/index.js --stdio

# Send an email to the candidate
echo '{"jsonrpc": "2.0","id": 5,"method": "tools/call","params": {"name": "contact_candidate", "args": {"subject": "Hello from AI!", "message": "This is a test email sent via the MCP server.", "reply_address": "[email protected]"}}}' | node dist/index.js --stdio

Each message must be on a single line with no line breaks within the JSON object.

Features

• Library-first design for integration into other applications
• Modular resource system for extending with custom candidate information
• TypeScript for type safety and better developer experience
• Implements the full Model Context Protocol specification
• Supports multiple transport types (STDIO, HTTP, Streamable HTTP)
• Minimal dependencies

Server Structure

src/
  ├── index.ts                # Main package entry point
  ├── server.ts               # MCP server factory with configuration
  ├── config.ts               # Configuration type definitions
  └── resources/              # Modular resource definitions
      └── index.ts            # Resource factory and implementation

MCP Protocol

This library implements the Model Context Protocol (MCP), a standardized way for LLMs to interact with external data and functionality. When integrated into your application, it exposes a stateless API that responds to JSON-RPC requests.

API Usage

Once integrated into your application, clients can interact with the MCP server by sending JSON-RPC requests. Here are examples of requests that your application would handle after integrating this library:

Initialize

curl -X POST http://your-application-url/mcp \
  -H "Content-Type: application/json" \
  -H "Accept: application/json" \
  -H "Accept: text/event-stream" \
  -d '{
    "jsonrpc": "2.0",
    "id": 1,
    "method": "initialize",
    "params": {
      "protocolVersion": "2024-11-05",
      "capabilities": {
        "roots": {
          "listChanged": true
        },
        "sampling": {}
      },
      "clientInfo": {
        "name": "ExampleClient",
        "version": "1.0.0"
      }
    }
  }'

Access Candidate Resources

curl -X POST http://your-application-url/mcp \
  -H "Content-Type: application/json" \
  -H "Accept: application/json" \
  -H "Accept: text/event-stream" \
  -d '{
    "jsonrpc": "2.0",
    "method": "resources/read",
    "params": {
      "uri": "candidate-info://resume-text"
    },
    "id": 2
  }'

Extending the Library

This library is designed to be extended with custom resources, tools, and prompts. Here's how to add your own resources:

import { McpServer, Resource } from '@jhgaylor/candidate-mcp-server';

// Create your custom resource class
class CustomCandidateResource extends Resource {
  constructor(candidateConfig) {
    super(
      `${candidateConfig.name} Custom Data`, 
      "candidate-info://custom-data", 
      async () => {
        return {
          contents: [
            { 
              uri: "candidate-info://custom-data", 
              mimeType: "text/plain", 
              text: "Your custom candidate data here"
            }
          ]
        };
      }
    );
  }
}

// Create server with standard configuration
const server = createServer(serverConfig, candidateConfig);

// Add your custom resource
const customResource = new CustomCandidateResource(candidateConfig);
customResource.bind(server);

// Connect with preferred transport
// ...

Adding Custom Tools

You can also extend the library with custom tools:

import { McpServer } from '@modelcontextprotocol/sdk/server/mcp.js';
import { z } from 'zod';
import { createServer } from '@jhgaylor/candidate-mcp-server';

// Create server with standard configuration
const server = createServer(serverConfig, candidateConfig);

// Add a custom tool
server.tool(
  'get_candidate_skills',
  'Returns a list of the candidate skills',
  {},
  async (_args, _extra) => {
    return {
      content: [
        { 
          type: "text", 
          text: "JavaScript, TypeScript, React, Node.js, MCP Protocol" 
        }
      ]
    };
  }
);

// Connect with preferred transport
// ...

Requirements

• Node.js 20+
• npm or yarn

License

MIT

Publishing to npm

Log in to npm if you haven't already:

npm login

Publish the package to npm (will run your prepublishOnly build):

npm publish

To bump, tag, and push a new version:

npm version patch    # or minor, major
git push origin main --tags