Video Commerce Intelligence MCP

AI-powered commercial intelligence from YouTube videos. Extract entities, score monetization opportunities, analyze audience intent, and discover market gaps -- all via the Model Context Protocol.

Give it a YouTube URL. It tells you everything commercially interesting about it -- and what to create next.

Quick Start

# Run directly (stdio transport, for local MCP use)
npx video-commerce-mcp

# Run as SSE server (for remote deployment)
npx video-commerce-mcp --transport sse --port 3001

Requires:

• OPENAI_API_KEY environment variable (GPT-4o-mini for entity extraction)

• Optional: Python 3 with youtube-transcript-api (pip install youtube-transcript-api) for reliable transcript fetching. Falls back to npm-based fetching if not available.

Related MCP server: YTPipe

Add to Claude Desktop

Add to your claude_desktop_config.json:

{
  "mcpServers": {
    "video-commerce": {
      "command": "npx",
      "args": ["video-commerce-mcp"],
      "env": {
        "OPENAI_API_KEY": "your-openai-api-key"
      }
    }
  }
}

Add to Claude Code

Create or edit .claude/mcp.json in your project root:

{
  "mcpServers": {
    "video-commerce": {
      "command": "npx",
      "args": ["video-commerce-mcp"],
      "env": {
        "OPENAI_API_KEY": "your-openai-api-key"
      }
    }
  }
}

Or connect to a remote SSE server:

{
  "mcpServers": {
    "video-commerce": {
      "type": "sse",
      "url": "https://your-server.example.com/sse"
    }
  }
}

Tools (12)

Layer 1 -- Video Intelligence

Layer 2 -- Market Intelligence

Pricing

Free tier: 5 calls/day (any tool) without payment, for testing and evaluation.

Paid tier: x402 micropayments in USDC on Base network. See the pricing column above for per-tool costs.

API key auth: Alternatively, configure API keys for authenticated access without x402.

Example Usage

analyze_video

Input:

{
  "youtube_url": "https://www.youtube.com/watch?v=abc123",
  "analysis_depth": "standard",
  "focus": ["entities", "monetization", "audience"]
}

Output (abbreviated):

{
  "video_id": "abc123",
  "title": "See This Chef's Amazing Kitchen Garden",
  "commercial_intent_score": 82,
  "entities": [
    {
      "name": "Helenium 'Sahin's Early Flowerer'",
      "category": "plant",
      "confidence": 0.94,
      "is_shoppable": true,
      "monetization_potential": {
        "affiliate_score": 0.85,
        "course_relevance": 0.6
      }
    }
  ],
  "audience_intent": {
    "dominant_intent": "seasonal_action",
    "intents": [{ "type": "seasonal_action", "score": 0.89 }]
  },
  "monetization": {
    "opportunities": [
      { "strategy": "affiliate_commerce", "score": 0.87 }
    ]
  }
}

discover_content_gaps

Input:

{
  "category": "autumn perennials",
  "region": "UK"
}

Output (abbreviated):

{
  "gaps": [
    {
      "topic": "helenium variety comparison",
      "demand_score": 0.78,
      "competition": 0.23,
      "opportunity_score": 0.85,
      "recommendation": "invest_now"
    }
  ],
  "emerging_topics": ["no-dig perennial borders"],
  "declining_topics": ["traditional herbaceous border maintenance"]
}

Remote Deployment (SSE)

# Start SSE server
npx video-commerce-mcp --transport sse --port 3001

# Health check
curl http://localhost:3001/health

Docker:

docker build -t video-commerce-mcp .
docker run -p 3001:3001 -e OPENAI_API_KEY=sk-... video-commerce-mcp

Configuration

Copy .env.example to .env and fill in your values. See the file for all available options.

Required:

• OPENAI_API_KEY -- OpenAI API key for GPT-4o-mini entity extraction

Optional:

• X402_ENABLED / X402_WALLET_ADDRESS -- Enable x402 micropayments

• API_KEYS -- Comma-separated API keys for authenticated access

• FREE_TIER_DAILY_LIMIT -- Free calls per day (default: 5)

• ANALYSIS_CACHE_DIR -- Cache directory (default: ~/.video-commerce-mcp/)

Programmatic Usage

import { createServer, startStdioServer } from "video-commerce-mcp";

// Use the server factory
const server = createServer();

// Or start directly
await startStdioServer();

Domain Expansion

The server is built on a domain-agnostic architecture. While the default vertical is gardening, the same pipeline works for:

• Cooking -- ingredients, equipment, techniques, cuisine styles

• DIY / Home improvement -- tools, materials, techniques, project types

• Tech reviews -- products, specs, alternatives, price points

• Fashion / Beauty -- products, brands, styles, occasions

• Fitness -- equipment, exercises, programs, supplements

Each vertical needs a domain dictionary, category keywords, and prompt tuning. The MCP framework stays identical.

See docs/verticals.md for implementation details.

OpenClaw Integration

Running OpenClaw for content production? Install this skill from ClawHub:

clawhub install video-commerce-intelligence

Or wire it directly via McPorter:

mcporter add video-commerce-mcp

Content team workflows

After each episode drops:

"Analyze this week's episode and give me affiliate links for the show notes: https://youtu.be/abc123"

The agent calls get_commercial_entities, then scan_affiliate_programs for the top entities, and returns a formatted list ready to paste into your CMS.

Planning next episode:

"What should we create next based on viewer demand in the startup tools space?"

The agent calls discover_content_gaps + track_category_lifecycle and returns the top 3 opportunities ranked by demand score and competition level.

Seasonal calendar:

"What's coming up in the next 90 days that our audience will care about?"

The agent calls get_seasonal_calendar for your region and returns upcoming events with demand multipliers.

OpenClaw agent config (direct MCP wiring)

mcpServers:
  - name: video-commerce
    command: npx video-commerce-mcp
    env:
      OPENAI_API_KEY: "${OPENAI_API_KEY}"
      MCP_API_KEYS: "${YOUR_API_KEY}"

Architecture

AI Agent (Claude, GPT, etc.)
     |
     | MCP Protocol (stdio or SSE)
     | x402 Payment Header (optional)
     v
Video Commerce Intelligence MCP
     |
     +-- Transcript Pipeline (fetch, preprocess, reduce tokens 70-90%)
     +-- NER Pipeline (extract, resolve, disambiguate, calibrate)
     +-- AI Orchestration (GPT-4o-mini, budget-managed)
     +-- Intelligence (audience intent, skills, quality, seasonal)
     +-- Market Intelligence (convergence, affiliates, authority, lifecycle)
     +-- Analysis Cache (SQLite, 7-day TTL)
     +-- Payment / Metering (x402, API key, free tier)

License

MIT