Salesforce MCP Server

This server provides MCP (Model Context Protocol) tools for interacting with Salesforce using the sf CLI. It exposes Salesforce data and metadata through standardized MCP tools that can be used with Claude Desktop and other MCP clients.

Features

• Query Tool (salesforce_query) - Run SOQL queries and return structured results
• List Objects (salesforce_list_objects) - Get all Salesforce object names in your org
• Describe Objects (salesforce_describe) - Get detailed field information for any Salesforce object
• List Flows (salesforce_list_flows) - Get all Salesforce flows with status and version information
• Describe Flow (salesforce_describe_flow) - Get the complete XML metadata for a specific flow
• List Reports (salesforce_list_reports) - Get all Salesforce reports with folder and usage information
• List Dashboards (salesforce_list_dashboards) - Get all Salesforce dashboards with folder and usage information

Prerequisites

• macOS (this setup is optimized for macOS)
• Python 3.11+

Installation

One-command setup:

git clone https://github.com/yourusername/sfmcp.git
cd sfmcp
./install.py  # Installs everything and configures Claude Desktop

The installer will automatically:

• Install Poetry and Python dependencies
• Install Node.js Salesforce CLI (if needed)
• Authenticate with your Salesforce org (or select existing auth)
• Create .env configuration
• Update Claude Desktop configuration

That's it! The script handles everything needed to get you up and running.

Manual Setup (Alternative)

If you prefer to set up manually or the automatic installer doesn't work:

1. Install dependencies:

   # Install Poetry (if needed)
   curl -sSL https://install.python-poetry.org | python3 -
   
   # Install Python dependencies
   poetry install
   
   # Install Salesforce CLI (if needed)
   npm install -g @salesforce/cli
   

2. Authenticate with Salesforce:

   sf org login web
   

3. Configure Claude Desktop by adding this to your claude_desktop_config.json:

   {
     "mcpServers": {
       "sfmcp": {
         "command": "/path/to/your/sfmcp/run-mcp.sh",
         "args": []
       }
     }
   }
   

4. Restart Claude Desktop to load the new MCP server.

Usage

Once configured, you can use these tools in Claude Desktop conversations:

• "List all Salesforce objects" - Uses salesforce_list_objects
• "Describe the Account object" - Uses salesforce_describe
• "Query all active accounts" - Uses salesforce_query
• "List all flows" - Uses salesforce_list_flows
• "Describe the Contact_Last_Reply_Date flow" - Uses salesforce_describe_flow
• "List all reports" - Uses salesforce_list_reports
• "List all dashboards" - Uses salesforce_list_dashboards

Example queries:

• SELECT Id, Name FROM Account LIMIT 10
• SELECT Id, Email FROM Contact WHERE Email != null

Development

Running the Server

STDIO Mode (for MCP clients):

poetry run sfmcp-stdio
# or
./scripts/run_stdio.sh

HTTP Mode (for testing):

poetry run sfmcp-http
# or
./scripts/run_http.sh

Code Quality

# Type checking
poetry run mypy sfmcp/

# Linting
poetry run ruff check sfmcp/

# Formatting
poetry run ruff format sfmcp/

Configuration

The server uses environment variables from .env file:

• SF_INSTANCE_URL - Your Salesforce instance URL
• SF_ACCESS_TOKEN - Salesforce access token (automatically managed)
• SF_ORG_ALIAS - Alias for your Salesforce org
• SF_USERNAME - Your Salesforce username
• SFMCP_HTTP_HOST - HTTP server host (default: 127.0.0.1)
• SFMCP_HTTP_PORT - HTTP server port (default: 3333)

Troubleshooting

"Salesforce CLI not found"

• Re-run the installer: python install.py
• Or install manually: npm install -g @salesforce/cli

"Authentication failed"

• Run sf org list to check authenticated orgs
• Re-run python install.py to set up authentication

"Permission denied"

• Make sure you have proper Salesforce org permissions
• Check that your access token hasn't expired