Windsurf

Skeet integrates seamlessly with Windsurf via Model Context Protocol (MCP), enabling your AI productivity tool to connect with your favorite development tools and services.

What is Model Context Protocol?

Model Context Protocol (MCP) is the technology that connects AI applications like Windsurf to external tools and services through Skeet. MCP allows:

• Secure connections between your AI tools and external services
• Context sharing across multiple applications
• Tool augmentation giving AI models the ability to perform actions in the real world
• Enhanced capabilities beyond what's built into the AI application itself

With MCP, your Windsurf assistant can interact with services like GitHub, Linear, Jira, and more, making it much more powerful and useful for your development workflow.

Setting Up Windsurf with Skeet

To use Skeet with Windsurf, you'll need to:

1. Get your Skeet MCP key from the Skeet dashboard
2. Configure Windsurf to use the Skeet MCP server
3. Connect integrations in your Skeet dashboard

Configuring MCP

To set up MCP with Windsurf, follow these steps:

1. Open Windsurf
2. Navigate to Settings > Advanced Settings (or use Command Palette > Open Windsurf Settings Page)
3. Scroll down to the Cascade section
4. Look for the MCP servers configuration section

This is where you'll add the Skeet MCP server to enable integrations.

Adding a New Server

The easiest way to add Skeet as an MCP server is through the Windsurf UI:

1. In the Cascade settings section, click the "Add Server" button
2. Choose "Add custom server +"
3. Configure the server with the following details:
   • Name: Skeet
   • URL: https://skeet.sh/ot/<YOUR_SKEET_MCP_KEY>
4. Click "Save" to add the server
5. Press the refresh button to update your available MCP servers

Remember to replace <YOUR_SKEET_MCP_KEY> with your actual Skeet MCP key from the dashboard.

Manual Configuration

For advanced users, you can directly edit the mcp_config.json file:

1. Locate the file at ~/.codeium/windsurf/mcp_config.json
2. Add the Skeet server to the configuration:

{
  "mcpServers": {
    "skeet.build": {
      "command": "npx",
      "args": [
        "-y",
        "supergateway",
        "--sse",
        "https://skeet.sh/ot/YOUR_SKEET_MCP_KEY"
      ]
    }
  }
}

3. Save the file and restart Windsurf for the changes to take effect

Verifying Your Connection

After setting up your MCP connection, you should verify it's working correctly:

1. Open Windsurf
2. Open Cascade (Windsurf's AI assistant)
3. Ask to list your Skeet integrations:

List my Skeet integrations

If the connection is successful, you'll see a response showing available integrations and their connection status.

Best Practices

Here are some best practices for using Skeet with Windsurf:

• Connect multiple integrations to maximize the utility of your AI assistant
• Use clear, specific prompts when asking Cascade to interact with external services
• Verify actions before letting the AI execute potentially destructive operations
• Maintain security by rotating your MCP key periodically
• Keep Windsurf updated to ensure compatibility with the latest MCP features

Example Use Cases

GitHub Integration

Create a GitHub issue in owner/repo with title "Fix navigation bug in header" and body "The dropdown menu in the navigation header doesn't close when clicking outside."

Linear Integration

Create a Linear issue with title "Implement user profile page" and priority high

Jira Integration

Create a Jira issue in project KEY with title "Refactor authentication service" and type Task

Integrations

Skeet supports a growing ecosystem of integrations that can be used with Windsurf:

Connect integrations in your Skeet dashboard to enable them for use with Windsurf.

Troubleshooting

If you encounter issues with your Skeet Windsurf integration:

Common Problems

• Cascade doesn't recognize MCP commands: Verify your MCP server is properly configured
• Authentication errors: Check that your MCP key is correct and not expired
• Integration not working: Ensure you've authorized the integration in the Skeet dashboard
• Tools not appearing: Remember that only tool endpoints are supported, not prompts or resources

Known Limitations

• MCP tool calls will consume credits regardless of success or failure
• Only tools are supported, not prompts or resources
• Tools that output images are not supported
• Only servers using the stdio transport type are supported

Getting Help

If you continue to experience issues:

• Review the MCP documentation for more information
• Review the Windsurf documentation for MCP-related guidance
• Contact Skeet Support with details about your problem