Salesforce Marketing Cloud MCP

Salesforce Marketing Cloud MCP Server

Skill by ara.so — Marketing Skills collection.

Overview

The Salesforce Marketing Cloud MCP Integration is a Model Context Protocol (MCP) server that enables AI agents (like Claude, GPT, or any LLM) to interact with Salesforce Marketing Cloud (SFMC) through natural language. It provides three core capabilities:

1. Query Data Extensions: Safely retrieve subscriber and customer data from SFMC Data Extensions
2. Trigger Journeys: Enroll contacts into Journey Builder workflows programmatically
3. Manage Subscribers: View, update, or unsubscribe contacts while respecting communication preferences

This server abstracts away the complexity of SFMC's SOAP/REST APIs, rate limits, and XML envelopes, providing LLMs with deterministic, safe tools for marketing automation.

Installation

Deploy to Vinkius Edge (Recommended)

The fastest way to deploy this MCP server is via Vinkius Edge, which provides automatic scaling, credential management, and enterprise compliance:

npx mcpfusion deploy

After deployment, Vinkius provides an endpoint URL that you connect to your LLM client (Claude Desktop, Cursor, etc.).

Local Development Setup

For local development and testing:

# Clone the repository
git clone https://github.com/vinkius-labs/salesforce-marketing-mcp.git
cd salesforce-marketing-mcp

# Install dependencies
npm install

# Run in development mode
npm run dev

Docker Deployment

docker pull vinkius/salesforce-marketing-mcp

docker run -d \
  -e SFMC_CLIENT_ID=$SFMC_CLIENT_ID \
  -e SFMC_CLIENT_SECRET=$SFMC_CLIENT_SECRET \
  -e SFMC_SUBDOMAIN=$SFMC_SUBDOMAIN \
  -e SFMC_ACCOUNT_ID=$SFMC_ACCOUNT_ID \
  -p 3000:3000 \
  vinkius/salesforce-marketing-mcp

Configuration

Environment Variables

The MCP server requires the following Salesforce Marketing Cloud credentials:

# Required: Your SFMC API credentials
SFMC_CLIENT_ID=your_client_id
SFMC_CLIENT_SECRET=your_client_secret

# Required: Your SFMC subdomain (e.g., mc123456789)
SFMC_SUBDOMAIN=your_subdomain

# Required: Your SFMC account MID
SFMC_ACCOUNT_ID=your_account_id

# Optional: Custom API base URL (defaults to REST endpoint)
SFMC_API_BASE_URL=https://your_subdomain.rest.marketingcloudapis.com

Obtaining SFMC Credentials

1. Log into Salesforce Marketing Cloud
2. Navigate to Setup → Apps → Installed Packages
3. Create a new package or select an existing one
4. Add a new API Integration component
5. Set the required scopes:
   • Data Extensions: Read, Write
   • Journeys: Execute, Read
   • List and Subscribers: Read, Write
6. Copy the Client ID and Client Secret
7. Note your Subdomain from the Marketing Cloud URL (e.g., mc123456789.exacttarget.com)
8. Find your Account MID in Setup → Account Settings

Claude Desktop Configuration

Add to your claude_desktop_config.json:

{
  "mcpServers": {
    "salesforce-marketing": {
      "command": "node",
      "args": ["/path/to/salesforce-marketing-mcp/build/index.js"],
      "env": {
        "SFMC_CLIENT_ID": "your_client_id",
        "SFMC_CLIENT_SECRET": "your_client_secret",
        "SFMC_SUBDOMAIN": "mc123456789",
        "SFMC_ACCOUNT_ID": "your_account_id"
      }
    }
  }
}

Available MCP Tools

1. query_data_extension

Query subscriber data from a Salesforce Marketing Cloud Data Extension.

Parameters:

• dataExtensionKey (string, required): The external key of the Data Extension
• fields (array, optional): Specific fields to retrieve (defaults to all)
• filter (object, optional): Filter criteria { field, operator, value }
• limit (number, optional): Maximum records to return (default: 100)

Example Usage:

// In your MCP client or AI agent context
{
  "tool": "query_data_extension",
  "arguments": {
    "dataExtensionKey": "CustomerPreferences",
    "fields": ["EmailAddress", "FirstName", "PreferredCategory"],
    "filter": {
      "field": "PreferredCategory",
      "operator": "equals",
      "value": "Electronics"
    },
    "limit": 50
  }
}

Natural Language Prompt:

"Show me the first 50 customers from the CustomerPreferences data extension who prefer Electronics"

2. trigger_journey

Enroll a contact into a Journey Builder workflow.

Parameters:

• journeyKey (string, required): The unique key of the Journey
• contactKey (string, required): Unique identifier for the contact
• eventData (object, optional): Additional event data to pass into the Journey

Example Usage:

{
  "tool": "trigger_journey",
  "arguments": {
    "journeyKey": "welcome-series-2024",
    "contactKey": "user@example.com",
    "eventData": {
      "RegistrationDate": "2024-01-15",
      "Source": "WebsiteSignup",
      "InterestCategory": "AI Tools"
    }
  }
}

Natural Language Prompt:

"Enroll user@example.com into the welcome-series-2024 journey with registration date January 15, 2024"

3. manage_subscriber

View, update, or unsubscribe a Marketing Cloud subscriber.

Parameters:

• action (string, required): One of get, update, or unsubscribe
• subscriberKey (string, required): Unique subscriber identifier (usually email)
• attributes (object, optional): Attributes to update (for update action)

Example Usage:

// Get subscriber details
{
  "tool": "manage_subscriber",
  "arguments": {
    "action": "get",
    "subscriberKey": "user@example.com"
  }
}

// Update subscriber attributes
{
  "tool": "manage_subscriber",
  "arguments": {
    "action": "update",
    "subscriberKey": "user@example.com",
    "attributes": {
      "FirstName": "Jane",
      "Industry": "Technology",
      "CompanySize": "50-200"
    }
  }
}

// Unsubscribe a contact
{
  "tool": "manage_subscriber",
  "arguments": {
    "action": "unsubscribe",
    "subscriberKey": "user@example.com"
  }
}

Natural Language Prompts:

"Get the subscriber details for user@example.com" "Update Jane's industry to Technology and company size to 50-200" "Unsubscribe user@example.com from all communications"

Common Patterns

Pattern 1: Audience Segmentation for Campaign

// Agent workflow: Identify high-value customers for a targeted campaign

// Step 1: Query customers with high engagement scores
const highValueCustomers = await queryDataExtension({
  dataExtensionKey: "CustomerEngagement",
  fields: ["EmailAddress", "EngagementScore", "LastPurchaseDate"],
  filter: {
    field: "EngagementScore",
    operator: "greaterThan",
    value: "80"
  },
  limit: 500
});

// Step 2: Trigger a journey for each qualified customer
for (const customer of highValueCustomers.records) {
  await triggerJourney({
    journeyKey: "vip-exclusive-offer",
    contactKey: customer.EmailAddress,
    eventData: {
      EngagementScore: customer.EngagementScore,
      LastPurchase: customer.LastPurchaseDate
    }
  });
}

Pattern 2: Event-Driven Journey Enrollment

// When a user completes a specific action, enroll them in a nurture journey

async function enrollAfterWebinar(attendeeEmail: string, webinarTitle: string) {
  await triggerJourney({
    journeyKey: "post-webinar-nurture",
    contactKey: attendeeEmail,
    eventData: {
      WebinarTitle: webinarTitle,
      AttendedDate: new Date().toISOString(),
      FollowUpPriority: "High"
    }
  });
}

Pattern 3: Preference Management

// Update subscriber preferences based on website activity

async function updatePreferencesFromBehavior(email: string, pageViews: string[]) {
  // Determine interest categories from page views
  const interests = categorizeInterests(pageViews);
  
  await manageSubscriber({
    action: "update",
    subscriberKey: email,
    attributes: {
      InterestCategories: interests.join(","),
      LastActivityDate: new Date().toISOString(),
      PreferenceSource: "WebsiteBehavior"
    }
  });
}

Pattern 4: Compliance-Safe Unsubscribe

// Process unsubscribe requests with audit trail

async function processUnsubscribeRequest(email: string, reason?: string) {
  // Log the unsubscribe reason to a data extension
  await queryDataExtension({
    dataExtensionKey: "UnsubscribeLog",
    // This would be an upsert operation in practice
  });
  
  // Actually unsubscribe the contact
  await manageSubscriber({
    action: "unsubscribe",
    subscriberKey: email
  });
}

Troubleshooting

Authentication Failures

Error: 401 Unauthorized or Invalid client credentials

Solution: Verify your SFMC credentials:

• Double-check SFMC_CLIENT_ID and SFMC_CLIENT_SECRET
• Ensure the API Integration has not expired
• Confirm the required scopes are enabled in your Installed Package

# Test credentials manually
curl -X POST https://$SFMC_SUBDOMAIN.auth.marketingcloudapis.com/v2/token \
  -H "Content-Type: application/json" \
  -d "{\"grant_type\":\"client_credentials\",\"client_id\":\"$SFMC_CLIENT_ID\",\"client_secret\":\"$SFMC_CLIENT_SECRET\"}"

Data Extension Not Found

Error: Data Extension with key 'XYZ' not found

Solution:

• Verify the external key (not the name) of the Data Extension
• Check that the Data Extension exists in the correct Business Unit
• Ensure your API user has permission to access the Data Extension

Rate Limiting

Error: 429 Too Many Requests

Solution:

• Implement exponential backoff in your agent logic
• When deploying on Vinkius Edge, rate limiting is handled automatically
• For local deployments, add retry logic:

async function withRetry<T>(fn: () => Promise<T>, maxRetries = 3): Promise<T> {
  for (let i = 0; i < maxRetries; i++) {
    try {
      return await fn();
    } catch (error) {
      if (error.status === 429 && i < maxRetries - 1) {
        await new Promise(resolve => setTimeout(resolve, Math.pow(2, i) * 1000));
        continue;
      }
      throw error;
    }
  }
  throw new Error('Max retries exceeded');
}

Journey Not Triggering

Error: Contact enrolled but journey does not fire

Solution:

• Verify the Journey is in Running state (not Draft or Stopped)
• Check that the journeyKey matches the API Event Definition Key
• Ensure the contact meets the Journey entry criteria
• Review Journey Builder logs in SFMC for entry rejections

TypeScript Build Errors

Error: Module resolution or type errors during local development

Solution:

# Clean build artifacts
rm -rf build/ node_modules/

# Reinstall dependencies
npm install

# Rebuild
npm run build

Advanced Configuration

Custom Timeout Settings

For long-running queries, adjust timeout values:

// In your local deployment
export const CONFIG = {
  SFMC_TIMEOUT_MS: 30000, // 30 seconds
  MAX_RECORDS_PER_QUERY: 1000
};

Multi-Business Unit Support

When working with multiple SFMC Business Units:

# Set the parent Account MID
SFMC_ACCOUNT_ID=parent_mid

# In tool calls, specify the target BU
{
  "tool": "query_data_extension",
  "arguments": {
    "dataExtensionKey": "CustomerData",
    "businessUnitId": "child_bu_mid"
  }
}

Resources

• Official Documentation
• Docker Hub
• GitHub Repository
• Salesforce Marketing Cloud API Docs
• Model Context Protocol Specification