How to Connect the Dice MCP Server to Your AI Assistant

What is an MCP Server? (And Why Should You Care?) 

If you've been using AI assistants like Claude, ChatGPT, or Gemini, you've probably hit their limits: they can't access real-time data, they can't search specific databases, and they definitely can't help you search for jobs beyond general advice. 

Model Context Protocol (MCP) changes that. 

Think of MCP as a universal adapter—like USB-C for your AI assistant. It lets your AI connect directly to external data sources and tools. Instead of copying and pasting job listings into your AI chat, you can give your AI direct access to search databases, analyze opportunities, and even build custom automations. 

The Dice MCP Server gives your AI assistant direct, real-time access to Dice's tech-only job database. This means you can: 

• Search using natural language instead of guessing exact keywords 

• Apply sophisticated filters (location radius, remote/hybrid/on-site, visa sponsorship, posting date) 

• Build job search automations that work while you sleep 

• Access quality tech jobs from Dice's carefully curated database 

Another thing you’ll love: No Dice login required. Just connect the MCP server and start searching. 

Video Tutorial: A Simple Setup Guide 

Want to see how this works before diving into setup? Watch our quick walkthrough: 

In this 5-minute tutorial, we'll show you how to connect the Dice MCP Server to Claude and run your first job search using natural language. 

Choosing Your Platform 

The Dice MCP Server works with three major AI assistants. Here's how they compare: 

Feature	Claude	ChatGPT	Gemini CLI
Setup Difficulty	Easiest	Moderate	Technical
Interface	Web/Mobile/Desktop	Web Only	Terminal
Subscription Required	Yes (Pro/Team/Enterprise)	Yes (Plus/Pro/Business/Enterprise)	No*
Best For	Quick start, mobile access	Developer Mode power users	Developers, automation
Time to Setup	2-3 minutes	5-10 minutes	10-15 minutes

*Gemini CLI is free and open-source but requires a Google AI Studio API key. 

Our Recommendation: 

• New to MCP? Start with Claude—fastest setup, works on mobile 

• Want advanced control? ChatGPT Developer Mode offers powerful customization 

• Terminal-native workflow? Gemini CLI integrates into your dev environment 

 

Platform-Specific Setup Guides 

Choose your platform below for step-by-step instructions: 

• Claude (Web, Mobile, & Desktop) 

• ChatGPT (Developer Mode) 

• Gemini CLI (Terminal) 

Setup Guide: Claude 

Requirements: 

• Claude Pro, Team, or Enterprise subscription 

• 2-3 minutes 

Available on: Web (claude.ai), Mobile (iOS/Android), Desktop app 

Quick Start (Web & Mobile) 

Step 1: Open the Connectors Directory 

From any chat: 

1. Click "Search and tools" (lower left of chat interface) 

2. Select "Add connectors" 

From Settings: 

1. Go to Settings → Connectors 

2. Click "Browse connectors" 

Step 2: Find and Connect Dice 

1. Search for "Dice" or browse Business tools / Developer tools 

2. Click the Dice MCP connector card 

3. Review capabilities:  

1. Natural language job search 

2. 13+ filter parameters 

3. Direct access to tech-only database 

4. Click "Connect" 

That's it—no login or permissions needed. 

Step 3: Enable in Conversations 

For each conversation where you want to use Dice: 

1. Click "Search and tools" in the chat interface 

2. Find Dice MCP in your tools list 

3. Toggle it ON 

Important: Tools are enabled per-conversation, so remember to turn on Dice MCP in each new chat. 

Step 4: Start Searching 

Try your first search: 

Find senior backend engineer roles in Denver
 

Or get more specific: 

Show me remote Python developer jobs posted in the last 3 days that offer visa sponsorship 
  

Claude will automatically use the Dice MCP connector and return structured results with direct application links. 

 

Desktop App Setup 

Claude Desktop lets you configure custom MCP servers for more control. 

Step 1: Locate Your Configuration File 

macOS: 

~/Library/Application Support/Claude/claude_desktop_config.json 

Windows: 

%APPDATA%\Claude\claude_desktop_config.json 

Linux: 

~/.config/Claude/claude_desktop_config.json 
  

Step 2: Edit the Configuration 

Open the file in any text editor. If it doesn't exist, create it. 

New file: 

{ 
  "mcpServers": { 
    "dice": { 
      "url": "https://mcp.dice.com/mcp", 
      "transport": "http" 
    } 
  } 
} 
  

Adding to existing servers: 

{ 
  "mcpServers": { 
    "existing-server": { 
      "url": "https://example.com/mcp", 
      "transport": "http" 
    }, 
    "dice": { 
      "url": "https://mcp.dice.com/mcp", 
      "transport": "http" 
    } 
  } 
} 
  

Step 3: Restart Claude Desktop 

Close the app completely and reopen. The Dice MCP server will load automatically. 

Step 4: Verify and Use 

Ask Claude: "Can you search the Dice job database?" 

Claude should confirm it has access to Dice tools. You're ready to search. 

 

Setup Guide: ChatGPT 

Requirements: 

• ChatGPT Plus, Pro, Business, Enterprise, or Education subscription 

• Developer Mode enabled 

• 5-10 minutes 

Available on: Web only (chat.openai.com) 

Understanding Developer Mode 

Developer Mode is ChatGPT's beta feature that provides full MCP support. It allows you to: 

• Connect custom MCP servers (like Dice) 

• Control tool permissions per conversation 

• Review and approve actions before execution 

Security Note: Developer Mode connects to external servers. Only connect trusted sources like Dice. 

Step 1: Enable Developer Mode 

For Individual Users (Plus/Pro) 

1. Open ChatGPT at chat.openai.com 

2. Click your profile icon (top-right) 

3. Select Settings 

4. Go to Apps & Connectors in the sidebar 

5. Scroll to Advanced settings 

6. Find Developer mode toggle 

7. Turn it ON 

8. Accept the security disclaimer 

9. Close Settings 

For Business/Enterprise/Education Users 

Your administrator must enable Developer Mode first: 

Administrators: 

1. Go to Workspace Settings 

2. Navigate to Permissions & Roles 

3. Find Connected Data section 

4. Enable Developer mode / Create custom MCP connectors 

Then individual users follow the steps above to activate it in their personal settings. 

Step 2: Add Dice MCP Connector 

Access Connector Settings 

From Settings: 

1. Profile icon → Settings 

2. Apps & Connectors 

3. Connectors section 

4. Click Create or Add connector 

From a conversation: 

1. Look for connector/tools menu 

2. Select Add connector 

Configure the Connector 

Enter the following: 

Connector Name: 

Dice Job Search 

Description: 

Search Dice's tech-only job database with natural language.  
Find roles by technology, location, workplace type, and more. 

MCP Server URL: 

https://mcp.dice.com/mcp 
  

Authentication: 

• Select No Authentication 

Transport Type: 

• Select SSE (Server-Sent Events) — recommended 

• Or HTTP Streaming (both work) 

Save and Verify 

1. Click Save or Create connector 

2. ChatGPT will connect to the Dice MCP server 

3. You should see the connector appear with search_jobs tool listed 

4. Connection successful! 

Step 3: Use Dice in Conversations 

Start a Developer Mode Chat 

1. Click the model selector at the top (Plus menu) 

2. Select Developer mode 

3. The interface switches to Developer Mode 

Add Dice to Your Conversation 

1. Click the Plus icon (+) in the message composer 

2. Select More 

3. Choose Dice Job Search 

4. The connector is now active 

Note: You must add connectors to each new Developer Mode conversation. 

Make Your First Search 

Try this: 

Find senior backend engineer jobs in Denver 

What happens: 

1. ChatGPT shows the tool call it will make 

2. You see search parameters 

3. Click Confirm to proceed 

4. Results appear with job titles, companies, and application links 

For more complex searches: 

Show me remote Python developer positions posted in the last 3 days that offer visa sponsorship 

ChatGPT will handle the filtering automatically. 

 

Setup Guide: Gemini CLI 

Requirements: 

• Gemini CLI installed 

• Command line familiarity 

• Google AI Studio API key (free) 

• 10-15 minutes 

Available on: Terminal (macOS, Windows, Linux) 

Best for: Developers who want terminal-native job search and automation capabilities. 

Step 1: Install Gemini CLI 

If you don't have Gemini CLI yet: 

Using npm (recommended): 

npm install -g @google/generative-ai-cli 

Using Homebrew (macOS): 

brew install gemini-cli 

Verify installation: 

gemini --version 

You should see version 0.4.0 or higher. 

Already installed? Make sure you're up to date: 

npm update -g @google/generative-ai-cli 

Step 2: Create Configuration Directory 

macOS/Linux: 

mkdir -p ~/.gemini 
  

Windows (PowerShell): 

New-Item -ItemType Directory -Force -Path "$env:USERPROFILE\.gemini" 
  

Step 3: Configure Dice MCP Server 

Open the configuration file in your text editor: 

macOS/Linux: 

nano ~/.gemini/settings.json 
  

Or use VS Code: 

code ~/.gemini/settings.json 
  

Windows: 

notepad "$env:USERPROFILE\.gemini\settings.json" 
  

Add This Configuration 

If the file is new or empty: 

{ 
  "mcpServers": { 
    "dice": { 
      "url": "https://mcp.dice.com/mcp" 
    } 
  } 
} 
  

If you already have other MCP servers: 

{ 
  "mcpServers": { 
    "existing-server": { 
      "url": "https://example.com/mcp" 
    }, 
    "dice": { 
      "url": "https://mcp.dice.com/mcp" 
    } 
  } 
} 
  

Save the file: 

• nano: Ctrl+X, then Y, then Enter 

• VS Code: Ctrl+S (Windows/Linux) or Cmd+S (macOS) 

• notepad: File → Save 

Step 4: Verify Configuration 

Check your JSON is valid: 

Using jq: 

cat ~/.gemini/settings.json | jq . 
  

Using Python: 

python3 -m json.tool ~/.gemini/settings.json 
  

If you see formatted output with no errors, you're good. 

Step 5: Launch and Connect 

Start Gemini CLI: 

gemini 
  

Verify the connection: 

/mcp list 
  

You should see: 

=== MCP Servers === 
● dice - connected (1 tool) 
  Tools: 
    - search_jobs: Search Dice job database with advanced filters 
  

The ● dice - connected status confirms success. 

Step 6: Make Your First Search 

Try a simple search: 

Find senior backend engineer jobs in Denver 
  

First-time confirmation: 

Tool call requested: dice/search_jobs 
Parameters: 
  keyword: "senior backend engineer" 
  location: "Denver" 
 
Confirm? (y/n) 
  

Type y and press Enter. 

Gemini displays results with job details and application links. 

Troubleshooting Common Issues 

"Connector Not Found" or "Cannot Connect" 

Solution: 

• Verify you have a paid subscription (Claude/ChatGPT) or Gemini CLI installed 

• Check the MCP server URL is exactly: https://mcp.dice.com/mcp 

• Restart your application or terminal 

• For ChatGPT: Confirm Developer Mode is enabled 

• For Gemini CLI: Verify JSON configuration syntax 

Tool Not Responding in Conversation 

Solution: 

• Claude: Make sure Dice MCP is toggled ON in the current conversation 

• ChatGPT: Add the Dice connector to your Developer Mode chat 

• Gemini CLI: Run /mcp list to verify connection status 

• Try starting a fresh conversation 

Search Returns No Results 

Solution: 

• Use broader search terms (e.g., "software engineer" instead of "senior staff principal engineer III") 

• Remove overly restrictive filters 

• Check spelling of technologies and locations 

• Verify your location has active job postings 

Configuration File Errors (Desktop/CLI) 

Solution: 

• Validate JSON syntax using jsonlint.com 

• Ensure proper quotes, commas, and brackets 

• Verify file is saved in the correct location 

• Check file extension is .json (not .json.txt) 

Best Practices 

Writing Effective Search Prompts 

Be specific but natural: 

• Good: "Find senior backend engineer roles in Denver with 5+ years experience" 

• Bad:  "Find engineering jobs" 

Use plain English: 

• Good: "Show me Python jobs that don't require on-call work" 

• Bad: "python -on_call:true location:remote" 

Iterate and refine: 

• Start broad, then narrow with follow-up questions 

• Ask your AI to explain search parameters if results seem off 

Tips for Managing Your Searches 

One topic per conversation: Keep job searches separate from other tasks for cleaner context. 

Enable tools intentionally: Only activate the connectors you need for each conversation. 

Save useful patterns: Bookmark conversations with valuable search strategies. 

Privacy and Security 

What Dice MCP accesses: 

• ✅ Publicly available job listings only 

• ❌ NOT your personal Dice account (if you have one) 

• ❌ NOT your resume or profile (unless you have given the LLM access to your resume)

• ❌ NOT your application history 

What happens with your searches: 

• Queries are processed by your AI assistant according to their privacy policy 

• The Dice MCP server doesn't store your search history 

• You control when and how the tool is used 

Building Job Search Automations 

Once you're comfortable with basic searches, you can create sophisticated workflows: 

Morning Job Digest 

Every morning, search for [your criteria] and summarize new postings  
with direct application links 

Continuous Monitoring 

Check daily for new [specific role] positions and alert me if  
any match [detailed requirements] 

Skills Gap Analysis 

Search for [target role] and identify the most common required  
skills that I should learn 

Salary Benchmarking 

Compare compensation ranges for [role] across [multiple cities]  
and provide a summary report 

Market Trend Tracking 

Track which technologies are mentioned most frequently in  
[job category] postings over the past month 

 

Technical Reference 

MCP Server Details 

Endpoint: https://mcp.dice.com/mcp 

 Protocol: Model Context Protocol (MCP) 

 Authentication: None required 

 Primary Tool: search_jobs 

Available Search Parameters 

• keyword (required): Job title or search keywords 

• location: Geographic location (city, state, or "remote") 

• radius: Search radius from location 

• radius_unit: "mi" (miles) or "km" (kilometers) 

• workplace_types: "Remote", "On-Site", "Hybrid" 

• employment_types: "FULLTIME", "CONTRACTS", "PARTTIME", "THIRD_PARTY" 

• posted_date: "ONE" (1 day), "THREE" (3 days), "SEVEN" (7 days) 

• willing_to_sponsor: Boolean for visa sponsorship 

• easy_apply: Boolean for simplified application 

• jobs_per_page: Results per page (1-100) 

• page_number: Pagination support 

What You Get Back 

Each search returns: 

• Job titles and detailed descriptions 

• Company names and information 

• Location details and workplace type 

• Salary ranges (when available) 

• Required skills and technologies 

• Direct application links 

• Posting dates 

 

Getting Help 

Technical Documentation:

https://www.dice.com/about/mcp

Dice Support 

For job listings, platform questions, or MCP technical issues: 

• Website: dice.com/support 

• Email: support@dice.com 

Platform-Specific Support 

Claude: 

• Help Center: support.claude.com 

• Documentation: Claude MCP Connectors guide 

ChatGPT: 

• Help Center: help.openai.com 

• Community: community.openai.com 

Gemini CLI: 

• GitHub: github.com/google-gemini/gemini-cli 

• Documentation: Official Gemini CLI docs 

Your job search is smarter and faster! Take control with natural language search, sophisticated filters, and automation that works for you—not the other way around. 

Ready to get started? Pick your platform above and connect the Dice MCP Server in minutes.