YouTube intelligence inside your AI tools
Add BrightBean to Claude Desktop, Cursor, or any MCP client with one config block. Ask for content gaps, title scores, and channel benchmarks in plain English. No tab switching, no curl commands, no copy-pasting JSON.
The problem
YouTube data lives in the wrong place
Tab switching
Open YouTube. Open vidIQ. Open a spreadsheet. Copy data between tools. Every context switch pulls you away from the work you were actually doing.
Manual API calls
Write curl commands, parse JSON responses, format results into something readable. The friction kills the workflow before you get to the insight.
With BrightBean MCP
Ask Claude in natural language, get structured answers back. No tab switching, no manual API calls. YouTube data shows up where you already work.
See it in action
A cooking creator plans next week's content with Claude
Here's what a typical session looks like. One config block, four queries, and a full content plan without leaving the chat window.
Add the config (30 seconds)
Paste this JSON block into your Claude Desktop config file. The same block works in Cursor settings and Claude Code. Once it's there, Claude can call BrightBean whenever you ask a YouTube-related question.
No SDK to install. No package manager. No build step. Just one config block and you're connected.
Config file locations
-
Claude Desktop
claude_desktop_config.json -
Cursor
settings.json -
Claude Code
.claude/settings.json
{
"mcpServers": {
"brightbean": {
"url": "https://api.brightbean.xyz/mcp",
"transport": "sse"
}
}
}
Ask Claude to find content gaps
You
Find content gaps in home cooking with high demand and low competition
BrightBean API response
{
"content_gaps": [
{
"topic": "5-ingredient slow cooker meals under $3",
"demand_score": 87,
"competing_videos_above_10k_views": 3,
"opportunity_rating": "high"
},
{
"topic": "meal prep for one person (no waste)",
"demand_score": 79,
"competing_videos_above_10k_views": 5,
"opportunity_rating": "high"
},
{
"topic": "pantry-only dinners no fresh ingredients",
"demand_score": 72,
"competing_videos_above_10k_views": 2,
"opportunity_rating": "very_high"
}
],
"niche": "home cooking",
"videos_analyzed": 14720
}
Claude
I found 3 underserved topics in home cooking. The strongest opportunity is "pantry-only dinners" with only 2 competing videos above 10k views and a demand score of 72. The slow cooker meals topic has the highest demand (87) but slightly more competition. I'd start with pantry-only dinners for the best ratio of demand to competition.
Score title ideas without leaving the chat
You
Score these 3 titles for the cooking niche: "Pantry-Only Dinner in 20 Minutes", "I Made a Week of Dinners From My Pantry", "No Grocery Store Challenge: 7 Days of Pantry Meals"
BrightBean API response
// Title 1: "Pantry-Only Dinner in 20 Minutes"
{ "score": 52, "patterns": ["timeframe", "specificity"] }
// Title 2: "I Made a Week of Dinners From My Pantry"
{ "score": 71, "patterns": ["personal_pronoun", "experience", "timeframe"] }
// Title 3: "No Grocery Store Challenge: 7 Days of Pantry Meals"
{ "score": 78, "patterns": ["challenge", "number_list", "curiosity_gap", "specificity"] }
// Niche average for cooking: 48
Claude
Title 3 wins with a score of 78, hitting 4 patterns. The "challenge" framing and specific "7 Days" timeframe create a curiosity gap that outperforms the other two. Title 2 is solid at 71, but Title 1 is only slightly above the cooking niche average of 48. Go with the challenge format.
Benchmark a competitor
You
How does @BudgetBites compare to the cooking niche average?
BrightBean API response
{
"channel": "@BudgetBites",
"posting_frequency": "2.1 videos/week",
"niche_avg_frequency": "1.4 videos/week",
"title_score_avg": 61,
"niche_title_score_avg": 48,
"top_pattern": "under_price",
"weakness": "low use of curiosity gaps and challenge formats"
}
Claude
@BudgetBites posts more often than the niche average (2.1 vs 1.4 videos/week) and their title scores are above average at 61 vs 48. Their strongest pattern is price anchoring. The weakness is low use of curiosity gaps and challenge formats, which is exactly what your "No Grocery Store Challenge" title does well. You have a structural advantage there.
Getting started
Same config block, every tool
The MCP config is identical across Claude Desktop, Cursor, and Claude Code. Paste it into the right file and you're done.
Claude Desktop
Open your config file and paste the block below. On macOS it's in ~/Library/Application Support/Claude/
{
"mcpServers": {
"brightbean": {
"url": "https://api.brightbean.xyz/mcp",
"transport": "sse"
}
}
}
File: claude_desktop_config.json
Cursor
Go to Cursor Settings, find the MCP section, and add the same config block.
{
"mcpServers": {
"brightbean": {
"url": "https://api.brightbean.xyz/mcp",
"transport": "sse"
}
}
}
File: settings.json
Claude Code
Add the config to your project's settings file or your global Claude Code config.
{
"mcpServers": {
"brightbean": {
"url": "https://api.brightbean.xyz/mcp",
"transport": "sse"
}
}
}
File: .claude/settings.json
Compatible tools
/ Claude Code
client
Why BrightBean MCP
What you get out of the box
30-second setup
One JSON config block. No SDK, no package install, no build step.
Natural language
Ask for YouTube data like you'd ask a colleague. Claude handles the API calls.
All 5 endpoints
Content gaps, title scoring, thumbnails, hooks, and channel benchmarks.
Common questions about BrightBean MCP
Which AI tools support MCP? +
Claude Desktop, Claude Code, Cursor, Windsurf, and a growing list of MCP-compatible tools. Any tool that supports the Model Context Protocol can connect to BrightBean using the same config block.
Do I need an API key for MCP? +
Yes. Sign up for a free BrightBean API key (500 calls, no credit card). The MCP server authenticates using your key, so every tool call is tied to your account and plan limits.
Can I use MCP and the REST API together? +
Yes. Same API key, same endpoints, same data. MCP is a convenience layer for AI tools. The REST API is for direct programmatic access. Use whichever fits the task, or both.
How does Claude know when to call BrightBean? +
MCP exposes BrightBean's tools (content_gaps, score_title, score_thumbnail, score_hook, benchmark) to Claude. When you ask a YouTube-related question, Claude automatically selects the right tool and calls it. You don't need to specify which endpoint to use.
Is there a limit to how many queries I can make? +
Same limits as the REST API. Each MCP tool call counts as one API call against your plan. The free tier includes 500 calls. If you need more, paid plans start at $19/month for 3,000 calls.
What happens if I exceed my plan? +
BrightBean returns a clear error message and Claude will tell you that you've hit your limit. No surprise charges. You can upgrade your plan or wait for the next billing cycle to reset your quota.
Can I use BrightBean MCP for research? +
Absolutely. MCP works well for exploratory research sessions. Ask Claude to find content gaps across multiple niches, benchmark several competitors in sequence, or score a batch of title ideas. The conversational format makes it easy to iterate on results without writing any code.
YouTube intelligence in your AI tools.
Free tier: 500 API calls, no credit card required.