> ## Documentation Index
> Fetch the complete documentation index at: https://docs.goldsky.com/llms.txt
> Use this file to discover all available pages before exploring further.

# Goldsky MCP Server

> Connect Goldsky documentation to AI tools with our hosted MCP server

## About the Goldsky MCP server

The Model Context Protocol (MCP) is an open protocol that creates standardized connections between AI applications and external services. Goldsky provides an MCP server that allows AI tools like Claude, Cursor, and other MCP clients to search and access our documentation directly. Your MCP-enabled AI tools can search across all Goldsky products and features, making it easier to:

* Find relevant documentation while coding
* Get accurate answers about Goldsky's capabilities
* Access examples and best practices in context
* Navigate between related features across products

### How the Goldsky MCP server works

When an AI tool has the Goldsky MCP server connected, it can search our documentation during response generation:

* The AI proactively searches Goldsky docs when relevant to your question
* Searches span all four core Goldsky products (Subgraphs, Mirror, Turbo, and Compose)
* Real-time documentation access ensures up-to-date information
* Context-aware results help you find the right product and feature

## Access the Goldsky MCP server

The Goldsky MCP server is hosted at:

```
https://docs.goldsky.com/mcp
```

## Connect the Goldsky MCP server

Choose your preferred AI tool to get started with the Goldsky MCP server:

<Tabs>
  <Tab title="Cursor">
    To connect the Goldsky MCP server to Cursor:

    <Steps>
      <Step title="Open MCP settings">
        1. Use <kbd>Command</kbd> + <kbd>Shift</kbd> + <kbd>P</kbd> (<kbd>Ctrl</kbd> + <kbd>Shift</kbd> + <kbd>P</kbd> on Windows) to open the command palette.
        2. Search for "Open MCP settings".
        3. Select **Add custom MCP**. This opens the `mcp.json` file.
      </Step>

      <Step title="Configure the Goldsky MCP server">
        In `mcp.json`, add the Goldsky server:

        ```json theme={null}
        {
          "mcpServers": {
            "Goldsky": {
              "url": "https://docs.goldsky.com/mcp"
            }
          }
        }
        ```
      </Step>

      <Step title="Test the connection">
        In Cursor's chat, ask "What tools do you have available?" Cursor should show the Goldsky MCP server as an available tool.

        <Check>
          Try asking: "How do I deploy a subgraph on Goldsky?" or "What data sinks does Mirror support?"
        </Check>
      </Step>
    </Steps>

    See the [Cursor documentation](https://docs.cursor.com/en/context/mcp#installing-mcp-servers) for more details.
  </Tab>

  <Tab title="VS Code">
    To connect the Goldsky MCP server to VS Code:

    <Steps>
      <Step title="Create MCP configuration">
        1. Create a `.vscode/mcp.json` file in your project root.
        2. In `mcp.json`, configure the Goldsky server:

        ```json theme={null}
        {
          "servers": {
            "Goldsky": {
              "type": "http",
              "url": "https://docs.goldsky.com/mcp"
            }
          }
        }
        ```
      </Step>

      <Step title="Restart VS Code">
        Restart VS Code to load the MCP configuration.
      </Step>

      <Step title="Verify the connection">
        Use GitHub Copilot Chat and ask about Goldsky features. The AI should be able to access Goldsky documentation through the MCP server.

        <Check>
          Try asking: "Show me how to create a Mirror pipeline" or "What's the difference between Mirror and Turbo?"
        </Check>
      </Step>
    </Steps>

    See the [VS Code documentation](https://code.visualstudio.com/docs/copilot/chat/mcp-servers) for more details.
  </Tab>

  <Tab title="Claude">
    To use the Goldsky MCP server with Claude:

    <Steps>
      <Step title="Add the Goldsky MCP server to Claude">
        1. Navigate to the [Connectors](https://claude.ai/settings/connectors) page in Claude settings.
        2. Select **Add custom connector**.
        3. Add the Goldsky MCP server:
           * Name: `Goldsky`
           * URL: `https://docs.goldsky.com/mcp`
        4. Select **Add**.
      </Step>

      <Step title="Access the MCP server in your chat">
        1. When using Claude, select the attachments button (the plus icon).
        2. Select the Goldsky MCP server.
        3. Ask Claude questions about Goldsky.

        <Check>
          Try asking: "How does Goldsky's Turbo product differ from Mirror?" or "Show me examples of Compose task triggers"
        </Check>
      </Step>
    </Steps>

    See the [Model Context Protocol documentation](https://modelcontextprotocol.io/docs/tutorials/use-remote-mcp-server) for more details.
  </Tab>

  <Tab title="Claude Code">
    To use the Goldsky MCP server with Claude Code, run the following command:

    ```bash theme={null}
    claude mcp add --transport http Goldsky https://docs.goldsky.com/mcp
    ```

    Test the connection by running:

    ```bash theme={null}
    claude mcp list
    ```

    <Check>
      Verify that "Goldsky" appears in the list of available MCP servers.
    </Check>

    See the [Claude Code documentation](https://docs.anthropic.com/en/docs/claude-code/mcp#installing-mcp-servers) for more details.
  </Tab>
</Tabs>

## Agent skills for AI coding assistants

For deeper AI integration beyond documentation search, Goldsky provides [Agent Skills](/ai-agent-skills)—workflow-based skills that guide AI assistants through complex blockchain data tasks like deploying pipelines, managing secrets, and debugging issues.

See the [AI agent skills](/ai-agent-skills) page for installation instructions and available skills.

## Using the Goldsky MCP server effectively

### Best practices for AI-assisted development

Once connected, your AI assistant can help you with:

<AccordionGroup>
  <Accordion title="Product selection and architecture">
    **When to use**: Starting a new project or evaluating options

    Ask questions like:

    * "Should I use Subgraphs or Mirror for my NFT marketplace?"
    * "When should I choose Turbo over Mirror?"

    The AI will search across product documentation to provide comparative guidance.
  </Accordion>

  <Accordion title="Configuration and setup">
    **When to use**: Setting up pipelines, subgraphs, or compose apps

    Ask questions like:

    * "Show me a Mirror pipeline config for decoding Uniswap events"
    * "How do I configure a Turbo pipeline with TypeScript transforms?"
    * "What's the syntax for Compose task triggers on ERC-20 transfers?"

    The AI will find relevant configuration examples and syntax.
  </Accordion>

  <Accordion title="Troubleshooting and debugging">
    **When to use**: Fixing issues or understanding errors

    Ask questions like:

    * "My subgraph deployment failed with error X, what does this mean?"
    * "How do I debug a failing Mirror transform?"
    * "What are common issues with Compose task execution?"

    The AI will search documentation for error explanations and solutions. If it is unable to help, don't hesitate to reach out to our support, and we will continue to add debugging guidance to our documentation to continuously improve the performance of the MCP server. Mention in your email that you are using the MCP server for priority support.
  </Accordion>
</AccordionGroup>

### Product-specific search tips

<Tabs>
  <Tab title="Subgraphs">
    **Key topics to search:**

    * Deploying and managing subgraphs
    * GraphQL schema and queries
    * Instant subgraphs (no-code)
    * Cross-chain/multi-chain indexing
    * Webhooks and event notifications
    * Migration from The Graph or Alchemy

    **Example queries:**

    * "How do I create an instant subgraph?"
    * "Show me how to deploy a cross-chain subgraph"
    * "What's the webhook payload format?"
  </Tab>

  <Tab title="Mirror">
    **Key topics to search:**

    * Pipeline configuration syntax
    * Data sources (subgraphs, direct indexing)
    * Sinks (Postgres, ClickHouse, Kafka, webhooks, etc.)
    * SQL transforms and decoding functions
    * Event schemas (EVM and non-EVM)
    * Performance optimization

    **Example queries:**

    * "How do I decode ERC-721 transfer events?"
    * "Show me Mirror pipeline config for Kafka sink"
    * "What functions are available for SQL transforms?"
  </Tab>

  <Tab title="Turbo">
    **Key topics to search:**

    * Pipeline configuration syntax
    * TypeScript transform syntax
    * Data sources (EVM, Solana, Stellar)
    * Dynamic tables and schemas
    * HTTP handlers and webhooks
    * Job mode vs streaming mode

    **Example queries:**

    * "Show me TypeScript transform examples"
    * "How do I configure Solana source in Turbo?"
    * "What's the syntax for dynamic table definitions?"
  </Tab>

  <Tab title="Compose">
    **Key topics to search:**

    * Task authoring and structure
    * Task triggers (time-based, blockchain events)
    * Context functions (fetch, EVM, collections)
    * Environment variables and secrets
    * Deployment and monitoring
    * Package imports

    **Example queries:**

    * "How do I create a task triggered by contract events?"
    * "Show me how to use EVM context to read contract state"
    * "How do I deploy and monitor Compose apps?"
  </Tab>
</Tabs>

<Tip>
  **Tip: Be specific about products**: When asking questions, mention the specific Goldsky product (Subgraphs, Mirror, Turbo, or Compose) to get more targeted results.

  Example: "How do I configure a **Mirror** pipeline?" vs "How do I configure a pipeline?" (could refer to Mirror or Turbo)
</Tip>