Skip to content

MCP for Maverick SDK (Beta)

Overview

Our MCP server will enable you to connect AI coding assistants to the Maverick SDK.

The Everysight MCP (Model Context Protocol) Server gives AI coding assistants direct access to:

  • SDK Documentation - 30 searchable guides and tutorials
  • API Reference - 66 classes across 15 packages
  • Code Samples - 8 official samples from GitHub (Kotlin & Swift)
  • Code Generation - Generate Screen boilerplate code
  • Code Validation - Check SDK compatibility
  • Interaction Simulation - Understand touch, sensors, and events

No API key required - Just connect and start coding!

Quick Start

Test the Server

# Health check
curl https://mcp-sdk.everysight.com/health

# List available tools
curl https://mcp-sdk.everysight.com/mcp/tools/list

# Search documentation
curl -X POST -H "Content-Type: application/json" \
  -d '{"query": "screen lifecycle"}' \
  https://mcp-sdk.everysight.com/mcp/tools/search_sdk_docs/execute

Connecting to Cursor IDE

Cursor has built-in MCP support. Here's how to connect:

Step 1: Open MCP Settings

  1. Open Cursor
  2. Go to SettingsFeaturesMCP Servers
  3. Or press Cmd+Shift+P (Mac) / Ctrl+Shift+P (Windows) and search "MCP"

Step 2: Add the Server

Click "Add MCP Server" and enter:

{
  "everysight-maverick": {
    "url": "https://mcp-sdk.everysight.com/mcp",
    "transport": "http"
  }
}

Or add to your ~/.cursor/mcp.json file:

{
  "mcpServers": {
    "everysight-maverick": {
      "url": "https://mcp-sdk.everysight.com/mcp",
      "transport": "http"
    }
  }
}

Note: The URL must include /mcp at the end for the JSON-RPC protocol.

Step 3: Verify Connection

In Cursor chat, type:

@everysight-maverick list tools

You should see 6 available tools.

Step 4: Start Using It

Ask questions like:

  • "Search the Maverick SDK for how to handle touch events"
  • "Generate a Screen class for displaying cycling metrics"
  • "Get the AR demo sample code for iOS"
  • "What happens when the glasses disconnect?"

Connecting to VS Code

Option A: Using Continue Extension

Continue is an open-source AI coding assistant for VS Code with MCP support.

  1. Install the Continue extension from VS Code marketplace
  2. Open Continue settings (~/.continue/config.json)
  3. Add the MCP server:
{
  "mcpServers": [
    {
      "name": "everysight-maverick",
      "url": "https://mcp-sdk.everysight.com/mcp"
    }
  ]
}
  1. Restart VS Code
  2. Use Continue chat to access Maverick SDK tools

Option B: Using Cline Extension

Cline is another AI assistant with MCP support.

  1. Install Cline from VS Code marketplace
  2. Open Cline settings
  3. Add MCP server URL: https://mcp-sdk.everysight.com/mcp

Option C: Using GitHub Copilot with MCP

GitHub Copilot is adding MCP support. Check the MCP Registry for the latest integration instructions.


Connecting to Windsurf (Codeium)

Windsurf is Codeium's AI-native IDE with MCP support.

Step 1: Open MCP Settings

  1. Open Windsurf
  2. Go to SettingsMCP Servers
  3. Or use the command palette and search "MCP"

Step 2: Add Server Configuration

{
  "mcpServers": {
    "everysight-maverick": {
      "url": "https://mcp-sdk.everysight.com/mcp",
      "transport": "http"
    }
  }
}

Step 3: Restart Windsurf

The Maverick SDK tools will be available in Cascade (Windsurf's AI assistant).


Connecting to Zed Editor

Zed is a high-performance editor with MCP support.

Step 1: Edit Settings

Open Zed settings (~/.config/zed/settings.json on Linux/macOS):

Step 2: Add MCP Configuration

{
  "language_models": {
    "mcp_servers": {
      "everysight-maverick": {
        "url": "https://mcp-sdk.everysight.com/mcp",
        "transport": "http"
      }
    }
  }
}

Step 3: Restart Zed

Use the assistant panel to access Maverick SDK tools.


Connecting to Claude Desktop

Claude Desktop supports MCP servers natively.

Step 1: Edit Configuration

Open the Claude Desktop config file:

  • macOS: ~/Library/Application Support/Claude/claude_desktop_config.json
  • Windows: %APPDATA%\Claude\claude_desktop_config.json

Step 2: Add Server

{
  "mcpServers": {
    "everysight-maverick": {
      "url": "https://mcp-sdk.everysight.com/mcp",
      "transport": "http"
    }
  }
}

Note: The URL must include /mcp at the end for the JSON-RPC protocol.

Step 3: Restart Claude Desktop

The Maverick SDK tools will now be available in your conversations.


Connecting to Other Tools

Generic HTTP Integration

Any tool that supports MCP over HTTP can connect using:

Setting Value
MCP JSON-RPC Endpoint https://mcp-sdk.everysight.com/mcp
Transport HTTP
Authentication None required

Important: Use the /mcp endpoint for MCP protocol (JSON-RPC). The REST endpoints below are for direct API access.

Available Endpoints

MCP JSON-RPC Protocol (for AI tools like Cursor, Claude Desktop):

Endpoint Method Description
/mcp POST JSON-RPC endpoint (initialize, tools/list, tools/call, etc.)

REST API (for direct curl/HTTP access):

Endpoint Method Description
/health GET Health check
/mcp/initialize POST MCP handshake
/mcp/tools/list GET List all tools
/mcp/tools/{name}/execute POST Execute a tool
/mcp/resources GET List resources
/mcp/prompts GET List prompt templates

Available Tools

1. search_sdk_docs

Search SDK documentation, guides, and tutorials.

Input:

{
  "query": "screen lifecycle",
  "max_results": 5
}

Example questions: - "How do I handle the screen lifecycle?" - "What's the rendering pipeline?" - "How do I use Bluetooth communication?"


2. search_api_reference

Search API classes, interfaces, and methods.

Input:

{
  "query": "UIElement",
  "filters": {
    "type": "class",
    "package": "UIKit.widgets"
  }
}

Example questions: - "What methods does the Screen class have?" - "Find all interfaces in UIKit.app.interfaces" - "What's the Animator class API?"


3. generate_screen_boilerplate

Generate Kotlin or Swift Screen class code.

Input:

{
  "screen_name": "CyclingMetricsScreen",
  "platform": "android"
}

Example questions: - "Generate a Screen class for displaying navigation" - "Create an iOS Screen for showing heart rate"


4. validate_sdk_code

Validate code for SDK compatibility.

Input:

{
  "code": "class MyScreen : Screen() { ... }",
  "language": "kotlin"
}

Example questions: - "Is this code compatible with SDK 2.6.1?" - "Check my Screen implementation for issues"


5. simulate_device_interaction

Understand SDK callbacks for device events.

Input:

{
  "interaction_type": "tap",
  "platform": "android"
}

Supported interactions: - Touch: tap, forward, backward, long_tap (TouchDirection enum: tap, forward, backward, longTap) - Connection: disconnect, connect - Battery: battery_low, battery_charging - Sensors: sensor_inertial, sensor_ypr, sensor_quaternion - Display: display_off, display_on

Example questions: - "What happens when the user swipes forward?" - "How do I handle battery low events?" - "Show me sensor callback code"


6. get_code_sample

Get official working samples from GitHub.

Input:

{
  "sample": "ar",
  "platform": "ios"
}

Available samples:

Sample Description
helloworld Basic "Hello World" screen
uielements All UI elements demo
animations Animation examples
ar AR with 3D objects
sensors Inertial sensors (YPR, quaternion)
images Image handling
glassesevents Touch, battery, connection
ota Firmware updates

Example questions: - "Show me the AR demo code for Android" - "Get the animations sample in Swift"


Example Workflows

Building Your First App

  1. Get started:

    "Search the SDK docs for getting started"

  2. Generate boilerplate:

    "Generate a Screen class called WelcomeScreen for Android"

  3. Add UI elements:

    "Search the API for Text and Image classes"

  4. Handle touch:

    "Simulate what happens when user taps"

  5. Get a working example:

    "Get the helloworld sample code"


Building an AR App

  1. Understand AR concepts:

    "Search docs for AR screen and coordinate systems"

  2. Get API details:

    "Search API for ArScreen, ArWindow, ArModel"

  3. Get working code:

    "Get the AR demo sample for iOS"

  4. Understand head tracking:

    "Simulate YPR sensor data"


Handling Device Events

  1. Learn about events:

    "Search docs for glasses events"

  2. Simulate interactions:

    "What happens when glasses disconnect?" "How do I handle low battery?"

  3. Get sample code:

    "Get the glassesevents sample"


Rate Limits

The server has generous rate limits for active development:

Limit Value
Per Minute 60 requests
Per Hour 1,000 requests
Per Day 10,000 requests

These limits are per IP address. Normal development usage won't hit these limits.


Troubleshooting

"Connection refused" or timeout

  • Check your internet connection
  • Verify the MCP URL: https://mcp-sdk.everysight.com/mcp
  • Try the health endpoint: https://mcp-sdk.everysight.com/health

"Tool not found"

Available tools are: - search_sdk_docs - search_api_reference - generate_screen_boilerplate - validate_sdk_code - simulate_device_interaction - get_code_sample

"Rate limit exceeded" (429)

  • Wait a minute and try again
  • Per-minute limit: 60 requests
  • Per-hour limit: 1,000 requests

Red dot / "Unexpected content type" error

If you see errors like: - Unexpected content type: text/html - Invalid content type, expected "text/event-stream"

Solution: 1. Make sure your URL ends with /mcp: https://mcp-sdk.everysight.com/mcp 2. Refresh the MCP server connection in your IDE settings 3. Restart your IDE completely

Green dot but tool calls fail (HTTP 500)

If listing tools works but calling them fails: 1. This may be a temporary server issue - wait a minute and retry 2. Check if the server is healthy: curl https://mcp-sdk.everysight.com/health 3. Try refreshing the MCP connection in your IDE

Tool returns empty results

  • Try broader search terms
  • Check spelling
  • Use get_code_sample with {"sample": "unknown"} to see all samples

Support & Resources

  • SDK Documentation: https://everysight.github.io/maverick_docs/
  • GitHub Samples: https://github.com/everysight-maverick/samples
  • SDK Repository: https://github.com/everysight-maverick/sdk
  • MCP Specification: https://modelcontextprotocol.io/docs

Learn about the Android Libraries Setup