MCP Server & Video Funnel Workflow – Full Tutorial

---

1. What This Setup Does

This solution automates:

1. Lead capture → From forms or Google Sheets.
2. Personalized video script generation → Using OpenAI (ChatGPT).
3. Video creation → Using HeyGen API (via MCP server).
4. Video tracking & updates → Saving video_id and video_url into Google Sheets.
5. Video status checks → A scheduled n8n workflow updates Google Sheets when videos are ready.

---

2. MCP Server Overview

Your MCP server:

• Endpoints:
  • POST /create-video → Creates a personalized HeyGen video and returns: jsonCopyEdit{ "success": true, "video_id": "...", "video_url": "https://app.heygen.com/videos/...", "video_script": "Generated text...", "status": "Processing", "date": "..." }
  • GET /video-status/:id → Checks if the video is ready.
  • POST /ai-leads → Forwards leads to n8n.
  • POST /store-lead → Saves lead data (optional for future).
• Technologies:
  • Node.js with Express.
  • Uses video.js for HeyGen + OpenAI integration.
  • Hosted in Docker with a docker-compose.yml.

---

3. MCP Server Setup

Step 3.1 – Directory Structure

pgsqlCopyEdit/docker/n8n/mcp-server/
├── tools/
│   ├── content.js
│   ├── lead.js
│   └── video.js
├── index.js
├── package.json
└── Dockerfile

---

Step 3.2 – Key Files

index.js

Handles all endpoints and routes (/create-video, /video-status, etc.).

video.js

• Generates personalized scripts using OpenAI.
• Calls HeyGen API to create videos.
• Constructs the video_url using video_id.

---

Step 3.3 – Environment Variables

Add these to your docker-compose.yml for mcp_server:

yamlCopyEdit- OPENAI_API_KEY=your-openai-api-key
- HEYGEN_API_KEY=your-heygen-api-key
- HEYGEN_AVATAR_ID=Byron_Business_Front_2_public
- HEYGEN_VOICE_ID=your-voice-id
- MCP_PORT=3000

---

Step 3.4 – Restart MCP Server

After modifying index.js or video.js, restart:

bashCopyEditdocker restart mcp-server

(If new dependencies are added, rebuild with docker compose build mcp_server.)

---

4. n8n Workflow: Video Creation

4.1 Workflow Steps

1. Trigger (Google Sheets or Form Submission)
   Detects new leads.
2. HTTP Node – Generate Personalized Video via MCP
   • POST https://mcp.xtreamsolution.net/create-video
   • Body Parameters: jsonCopyEdit{ "name": "{{ $json['full name'] }}", "email": "{{ $json.email }}", "company": "{{ $json.company }}", "industry": "{{ $json.industry }}", "interested": "{{ $json['interested services'] }}" }
3. Append/Update Google Sheets
   • Map MCP response:
     • VideoID → {{ $json.video_id }}
     • VideoURL → {{ $json.video_url }}
     • Status → {{ $json.status }}
     • Script → {{ $json.video_script }}

---

5. n8n Workflow: Video Status Check

This is a separate workflow.

5.1 Steps

1. Scheduler Node
   Runs every 10 minutes.
2. Google Sheets Node (Read)
   Filters rows where status = Processing.
3. HTTP Node – Check Video Status
   Calls: bashCopyEditGET https://mcp.xtreamsolution.net/video-status/{{ $json.VideoID }}
4. Google Sheets Node (Update)
   Updates:
   • Status → Completed when HeyGen returns completed.
   • VideoURL remains as is.

---

6. Testing Your Setup

6.1 MCP Server Test

bashCopyEditcurl -X POST https://mcp.xtreamsolution.net/create-video \
-H "Content-Type: application/json" \
-d '{
  "name":"John Doe",
  "company":"Acme Corp",
  "industry":"Finance",
  "interested":"AI chatbots"
}'

You should see a response with video_id, video_url, etc.

---

6.2 n8n Test

• Execute the “Generate Personalized Video” node standalone.
• Check if the response matches what Google Sheets expects.

---

7. Best Practices

• Error Handling in MCP: Always return { success: false } with HTTP 200 instead of 500 to prevent n8n from halting.
• Google Sheet Columns: Keep columns for video_id, video_url, status, and script.
• Monitor Logs: Use docker logs -f mcp-server for debugging.
• Video URLs: Constructed as https://app.heygen.com/videos/{{ video_id }}.