Skip to main content
This guide walks you through connecting the ShingleAI MCP server to Windsurf, allowing Cascade (Windsurf’s AI) to access your contacts, messages, and other business data.

​
Prerequisites

Before you begin, ensure you have:

​
Configuration

Windsurf uses a configuration file that follows the same schema as Claude Desktop.

​
Configuration File Location

~/.codeium/windsurf/mcp_config.json

​
Step 1: Create the configuration file

mkdir -p ~/.codeium/windsurf
touch ~/.codeium/windsurf/mcp_config.json

​
Step 2: Add ShingleAI server

Add the following configuration to your mcp_config.json: 
{
    "mcpServers": {
        "shingleai": {
            "command": "npx",
            "args": [
                "mcp-remote",
                "https://mcp.shingleai.com/sse"
            ],
            "env": {
                "MCP_HEADERS": "Authorization: Bearer YOUR_MCP_KEY"
            }
        }
    }
}
Replace YOUR_MCP_KEY with your actual MCP key.
Never hardcode API keys in configuration files that might be shared. Use environment variables instead.
For better security, use environment variables: 
{
    "mcpServers": {
        "shingleai": {
            "command": "npx",
            "args": [
                "mcp-remote",
                "https://mcp.shingleai.com/sse"
            ],
            "env": {
                "MCP_HEADERS": "Authorization: Bearer ${SHINGLEAI_MCP_KEY}"
            }
        }
    }
}
Set the environment variable in your shell profile: 
export SHINGLEAI_MCP_KEY=mcp_live_abc123...

​
Step 4: Restart Windsurf

After saving your configuration, restart Windsurf for the changes to take effect.

​
Alternative: UI Configuration

You can also configure MCP servers through Windsurf’s settings:
  1. Open the Command Palette with Cmd+Shift+P (Mac) / Ctrl+Shift+P (Windows/Linux)
  2. Type “Open Windsurf Settings”
  3. Navigate to Cascade > Plugins
  4. Add ShingleAI as a custom MCP plugin

​
Verification

To verify the connection is working:
  1. Open Cascade (Windsurf’s AI assistant)
  2. Ask: “What ShingleAI tools do you have access to?”
  3. Cascade should list the available tools

​
Example Usage

Once connected, you can use ShingleAI tools in Cascade:
TaskExample Prompt
Find contact info”Look up contact information for our vendor”
Search communications”Find messages mentioning the deployment issue”
View thread”Show the full conversation thread with support”

​
Troubleshooting

Possible causes:
  • Configuration file in wrong location
  • Invalid JSON syntax
  • Windsurf not restarted
Solutions:
  1. Verify the file is at ~/.codeium/windsurf/mcp_config.json
  2. Validate your JSON syntax
  3. Completely restart Windsurf
Possible causes:
  • Invalid or expired MCP key
  • Environment variable not set
Solutions:
  1. Verify your MCP key in the ShingleAI dashboard
  2. Ensure the environment variable is exported
  3. Try launching Windsurf from terminal after exporting the variable
Cascade has a limit of 100 total tools. If you have many MCP servers configured, you may need to disable some tools.Solution:
  1. Open Windsurf Settings > Cascade > Plugins
  2. Navigate to the Tools tab
  3. Toggle off tools you don’t need
Windsurf supports both stdio and HTTP transports for MCP servers. ShingleAI uses SSE (via mcp-remote) for optimal performance.

​
Next Steps

Available Tools

Explore all available MCP tools

Authentication

Configure tool permissions