How to Connect YouTube Intelligence to Claude Desktop via MCP

BrightBean’s MCP integration lets you ask Claude questions about YouTube strategy and get data-backed answers from live API calls: Content gaps in your niche, scored title candidates, hook analysis, competitive benchmarks. Here’s how to set it up in under 5 minutes.

What is MCP?

Model Context Protocol is an open standard that lets AI assistants connect to external tools and data. Instead of relying only on training data, Claude can make live API calls during a conversation: Search the web, query databases, or call specialized APIs like BrightBean.

Think of it as giving Claude hands. Without MCP, Claude can only talk. With MCP, Claude can do things, like analyzing your YouTube niche, scoring title candidates, and benchmarking your channel against competitors.

What You’ll Need

• Claude Desktop (free download from claude.ai/download)
• BrightBean API key (free at brightbean.xyz — 500 calls, no credit card)

That’s it. No Python. No terminal. No code.

Step 1: Get Your BrightBean API Key

1. Go to brightbean.xyz
2. Sign up for a free account
3. Copy your API key from the dashboard (it starts with bb-)

Keep this key handy. You’ll paste it into the config file in the next step.

Step 2: Add the MCP Configuration

Open Claude Desktop’s settings and add BrightBean as an MCP server.

On macOS

Open the config file:

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

Or open it with any text editor. If the file doesn’t exist, create it.

On Windows

Open the config file at:

%APPDATA%\Claude\claude_desktop_config.json

The Configuration

Add BrightBean to the mcpServers section:

{
  "mcpServers": {
    "brightbean": {
      "command": "npx",
      "args": [
        "-y",
        "@brightbean/mcp-server"
      ],
      "env": {
        "BRIGHTBEAN_API_KEY": "bb-your-api-key-here"
      }
    }
  }
}

Replace bb-your-api-key-here with your actual API key.

If you already have other MCP servers configured, add brightbean alongside them:

{
  "mcpServers": {
    "some-other-server": {
      "command": "...",
      "args": ["..."]
    },
    "brightbean": {
      "command": "npx",
      "args": [
        "-y",
        "@brightbean/mcp-server"
      ],
      "env": {
        "BRIGHTBEAN_API_KEY": "bb-your-api-key-here"
      }
    }
  }
}

Save the file and restart Claude Desktop.

Step 3: Verify the Connection

After restarting Claude Desktop, you should see BrightBean listed as an available tool. Look for the tools icon (hammer) in the bottom of the chat input area.

To test, start a new conversation and ask:

“What content gaps exist in the home fitness niche?”

Claude should call BrightBean’s /content-gaps endpoint and return a list of underserved topics with demand scores, supply gaps, and opportunity ratings.

If you see real data in the response, you’re connected.

Step 4: Ask Claude About YouTube Strategy

Now the fun part. Here are example prompts that show what’s possible:

Find Content Opportunities

“I run a cooking channel focused on meal prep. What content gaps exist in my niche? Show me the top 10 opportunities.”

Claude will call /content-gaps and explain each opportunity in context: What the topic is, why it’s underserved, and what angle to take.

Score Title Candidates

"I’m making a video about meal prepping for the week. Which of these titles is best?

1. How to Meal Prep for the Entire Week in 2 Hours
2. I Meal Prepped for 30 Days Straight — Here’s What Happened
3. The Only Meal Prep Guide You’ll Ever Need"

Claude will score each title and explain why one outperforms the others, breaking down clarity, curiosity, keyword strength, and click probability.

Analyze Your Hook

“Here’s the opening for my next video: ‘Last month I spent $400 on takeout. This month I spent $45 on groceries and ate better than ever. Let me show you exactly how.’ Score this hook and suggest improvements.”

Claude calls /analyze/hook and returns the hook type, retention score, and specific suggestions for tightening the opening.

Benchmark Your Channel

“How does my channel ‘MealPrepMonday’ compare to similar channels in the meal prep niche? What are my strengths and weaknesses?”

Claude calls /benchmark and provides a competitive analysis: Where you’re outperforming, where you’re lagging, and what to prioritize.

Chain Multiple Endpoints

The real power comes from combining endpoints in a single conversation:

“Find the top 5 content gaps in the meal prep niche. For each one, generate 3 title candidates and score them. Then build me a 2-week content calendar with the winning titles.”

Claude will chain /content-gaps → /score/title (15 calls) → calendar assembly, giving you a complete content plan in a single response. This is the same workflow as the LangChain agent tutorial, but without writing any code.

Advanced: Multi-Turn Strategy Sessions

MCP conversations maintain context across messages, enabling deep strategy sessions:

Turn 1: “Find content gaps in the photography niche” Claude returns 10 opportunities

Turn 2: “Those first 3 look great. Generate title options for each.” Claude scores titles for the selected topics

Turn 3: “I like the titles for topics 1 and 3, but topic 2 needs something more curiosity-driven. Try again.” Claude generates and scores new options

Turn 4: “Now analyze this hook I wrote for topic 1: [hook text]” Claude scores the hook and suggests improvements

Turn 5: “Put it all together into a content brief for topic 1.” Claude assembles the brief with gap data, winning title, scored hook, and content recommendations

This iterative, conversational workflow is impossible with a traditional API client. You’d need to write code for each step. With MCP, you just talk.

Works with Cursor, Windsurf, and Other MCP Clients

The same MCP configuration works in any MCP-compatible application:

Cursor

Add to .cursor/mcp.json in your project:

{
  "mcpServers": {
    "brightbean": {
      "command": "npx",
      "args": ["-y", "@brightbean/mcp-server"],
      "env": {
        "BRIGHTBEAN_API_KEY": "bb-your-api-key-here"
      }
    }
  }
}

Windsurf

Add to ~/.codeium/windsurf/mcp_config.json:

{
  "mcpServers": {
    "brightbean": {
      "command": "npx",
      "args": ["-y", "@brightbean/mcp-server"],
      "env": {
        "BRIGHTBEAN_API_KEY": "bb-your-api-key-here"
      }
    }
  }
}

VS Code with Continue

Add to your Continue config. The server definition is identical. Just place it in Continue’s MCP configuration section.

In all cases, you get the same BrightBean tools: Content gaps, title scoring, thumbnail scoring, hook analysis, and channel benchmarking, all accessible through your coding assistant’s natural language interface.

Use Cases for Developers

If you’re a developer, MCP + BrightBean is particularly powerful:

• Build and test prompts for YouTube content agents interactively before committing to code
• Explore the API without writing HTTP requests — just ask Claude to call endpoints
• Prototype workflows in conversation, then translate the working logic into code
• Debug integrations by asking Claude to call specific endpoints with specific parameters

It’s like having an interactive API playground built into your AI assistant.

Troubleshooting

“BrightBean tools not showing up”

1. Ensure the config file is saved in the correct location
2. Restart Claude Desktop completely (quit and reopen)
3. Check that your API key is correct (starts with bb-)
4. Verify that npx is available on your system (npx --version in terminal)

“API calls returning errors”

1. Check your API key hasn’t expired
2. Verify you haven’t exceeded the 500 calls free tier limit
3. Check the error message — BrightBean returns descriptive errors with fix suggestions

“Claude isn’t using BrightBean”

If Claude answers from general knowledge instead of calling BrightBean, try being explicit:

“Use BrightBean to find content gaps in the cooking niche”

Or reference a specific endpoint:

“Call BrightBean’s /score/title endpoint to score this title: [title]”

Related Reading

• How to Build a YouTube Content Planning Agent — The code-based version of this workflow using LangChain
• How to Run a YouTube Content Gap Analysis — Deep dive into interpreting content gap results
• YouTube Data API vs Intelligence API — Why MCP + BrightBean gives you more than the YouTube Data API

Try YouTube intelligence in Claude Desktop. Get your free BrightBean API key — 500 calls, no credit card required. Set up MCP in 5 minutes at brightbean.xyz.