[Beta] MCP for API docs and sample apps

Updated: May 24, 2026

Connect hosted Experiences SDK MCP servers in AI-powered IDEs like Cursor, Windsurf, and Claude Desktop so your coding assistant can query APIs, generate integration code, and search mobile sample repositories. Capabilities may change during the beta.

For Web SDK integration (npm MCP that analyzes your project and generates widget code), see Web SDK MCP for AI-assisted integration.

These servers allow your AI assistant to:

Query Experiences SDK APIs
Generate real integration code
Search full sample repositories
Troubleshoot implementation issues inside your IDE

Recommended workflow

If you're integrating the Experiences SDK, start with the sample apps.

They show working setup and end-to-end behavior, including common integration areas like widgets, deep links, push notifications, ads, and search. They're also a fast way to verify that your API key and environment configuration are correct before troubleshooting an issue in your own app.

Who should use MCP servers

This is most useful for developers who integrate the Experiences SDK on mobile or need API reference in the IDE.

App owners can also use the core docs MCP server to understand capabilities and integration requirements. You can connect it from any AI editor that supports MCP, like Cursor or Claude Desktop.

What you get

You can connect these hosted MCP servers:

1 core Experiences documentation server (API specs and code snippets)
3 sample app repository servers (Android, iOS, React Native)

Connect only the sample app MCP servers for the platforms you use.

For Web (React, Next.js, Vue, Angular, vanilla JavaScript), use the separate npm MCP in Web SDK MCP for AI-assisted integration. It runs locally and is tuned for Web SDK projects.

If you want the sample app links in one place (without MCP setup), see Sample apps (all platforms).

Available MCP servers

Experiences docs (core server)

The primary MCP server provides your AI assistant with direct access to the Experiences SDK APIs and documentation.

You can ask it to:

Explore SDK capabilities
"How do I configure a data source to load stories by label?"
Generate integration code
"Generate a Kotlin snippet that sets up a moments row widget with a custom layout preset."
Troubleshoot issues
"My widget is not loading content. What could be wrong?"

Recommended for all developers.

Core server tools

If your AI editor supports MCP tools, it can use these actions on the core server:

search_specs to search available APIs. For example, "What APIs are available for fetching stories?"
get_request_body to inspect request parameters. For example, "What parameters can I pass when requesting moments?"
get_response_schema to understand response models. For example, "What data model is returned when I fetch videos?"
get_code_snippet to generate integration code. For example, "Generate a Kotlin function to retrieve stories by label."
execute_request to test API calls. For example, "Fetch stories filtered by a specific label to verify my setup."

Sample app documentation servers

Each platform has a dedicated MCP server that exposes the full sample repository, including:

Documentation
Production-ready code
Implementation patterns
Best practices

Platform	Repository
Android	https://github.com/WSCSports/blaze-sample-android-v2
iOS	https://github.com/WSCSports/blaze-sample-ios-v2
React Native	https://github.com/WSCSports/blaze-sample-reactNative

With these connected, your AI assistant can:

Fetch documentation
"How do I initialize the SDK on Android?"
Search implementation examples
"Show me a widget implementation example for iOS."
Understand architectural patterns
"How is CTA handled in the React Native sample?"

Use sample apps even if you don't use MCP

You can use the sample apps without MCP:

Browse the repositories to learn implementation patterns.
Run the sample app to validate your API key, environment configuration, and basic widget setup.
Compare your implementation against a known working baseline when something doesn't behave as expected.

Setup per AI editor

Add the servers to your MCP configuration file.
Add 1-4 servers based on the platform you use.

Add to ~/.cursor/mcp.json:

{
  "mcpServers": {
    "wsc-blaze": {
      "url": "https://dev.wsc-sports.com/mcp"
    },
    "blaze-sample-android-v2 Docs": {
      "url": "https://gitmcp.io/WSCSports/blaze-sample-android-v2"
    },
    "blaze-sample-ios-v2 Docs": {
      "url": "https://gitmcp.io/WSCSports/blaze-sample-ios-v2"
    },
    "blaze-sample-reactNative Docs": {
      "url": "https://gitmcp.io/WSCSports/blaze-sample-reactNative"
    }
  }
}

Add to ~/.codeium/windsurf/mcp_config.json:

{
  "mcpServers": {
    "wsc-blaze": {
      "url": "https://dev.wsc-sports.com/mcp"
    },
    "blaze-sample-android-v2 Docs": {
      "url": "https://gitmcp.io/WSCSports/blaze-sample-android-v2"
    },
    "blaze-sample-ios-v2 Docs": {
      "url": "https://gitmcp.io/WSCSports/blaze-sample-ios-v2"
    },
    "blaze-sample-reactNative Docs": {
      "url": "https://gitmcp.io/WSCSports/blaze-sample-reactNative"
    }
  }
}

Add to claude_desktop_config.json:

{
  "mcpServers": {
    "wsc-blaze": {
      "url": "https://dev.wsc-sports.com/mcp"
    },
    "blaze-sample-android-v2 Docs": {
      "url": "https://gitmcp.io/WSCSports/blaze-sample-android-v2"
    },
    "blaze-sample-ios-v2 Docs": {
      "url": "https://gitmcp.io/WSCSports/blaze-sample-ios-v2"
    },
    "blaze-sample-reactNative Docs": {
      "url": "https://gitmcp.io/WSCSports/blaze-sample-reactNative"
    }
  }
}

Example

Ask your AI assistant to help you integrate SDK features using real sample code and documentation.

On Android, how do I set up a stories row widget with a custom layout and data source?

MCP Answer:

To set up a stories row widget, you need three things: a layout preset, a data source, and the initWidget() call.

Add the widget view to your XML layout.

<com.blaze.blazesdk.features.stories.widgets.row.BlazeStoriesWidgetRowView    
android:id="@+id/storiesRowWidgetView"    
android:layout_width="match_parent"
android:layout_height="160dp" />

Choose a layout preset and configure the data source.

// Pick a layout preset
val widgetLayout = BlazeWidgetLayout.Presets.StoriesWidget.Row.circles 

// Configure the data source with your label
val dataSource = BlazeDataSourceType.Labels(
  blazeWidgetLabel = BlazeWidgetLabel.singleLabel("your-label-here"),
  orderType = null
)

Initialize the widget.

binding.storiesRowWidgetView.initWidget(
  widgetLayout = widgetLayout,	
  dataSource = dataSource,
  widgetId = "stories-row-id",
  widgetDelegate = this,
  shouldOrderWidgetByReadStatus = true
)

Other available row presets include BlazeWidgetLayout.Presets.StoriesWidget.Row.verticalRectangles. You can customize dimensions, spacing, and item styles by modifying the layout object before passing it to initWidget().
For the full implementation, see StoriesRowFragment.kt in the sample app.

Getting started

Add the MCP server configuration to your editor (see Setup above).
Restart your editor or refresh the MCP connection.
Start a new AI chat and ask a question about the Experiences SDK.

Updated 3 months ago

Did this page help you?