MCP Tool Reference

The following tools are exposed by superglue’s MCP server. The input schemas are defined using Zod in mcp-server.ts.

​

1. superglue_execute_tool

• Description: Execute a specific Superglue tool by ID. Use this when you know the exact tool needed for a task.
• Input Schema: ExecuteToolInputSchema
  • id: The ID of the tool to execute.
  • payload: (Optional) JSON payload to pass to the tool.
  • credentials: (Optional) JSON credentials for the tool execution.
  • options: (Optional) Request configuration (caching, timeouts, retries, etc.).

• Example Usage (Conceptual MCP Call):

  // MCP callTool params
  {
    "toolName": "superglue_execute_tool",
    "inputs": {
      "id": "tool-id-123",
      "payload": { "inputData": "example" },
      "credentials": { "apiKey": "your-api-key" }
    }
  }

​

2. superglue_build_new_tool

• Description: Build a new integration tool from natural language instructions. Use when existing tools don’t meet requirements. Built tools are immediately saved and dynamically made available as new tools.
• Input Schema: BuildToolInputSchema
  • instruction: Natural language instruction for building the tool.
  • payload: (Optional) Example JSON payload for the tool. This should be data needed to fulfill the request (e.g. a list of ids to loop over), not settings or filters.
  • systems: Array of SystemInputSchema defining the systems the tool can interact with.
    • id: Unique identifier for the system.
    • urlHost: Base URL/hostname for the system.
    • urlPath: (Optional) Base path for API calls.
    • documentationUrl: (Optional) URL to API documentation.
    • credentials: (Optional) Credentials for accessing the system. MAKE SURE YOU INCLUDE ALL OF THEM BEFORE BUILDING THE CAPABILITY, OTHERWISE IT WILL FAIL.
  • responseSchema: (Optional) JSONSchema for the expected response structure.

• Example Usage (Conceptual MCP Call):

  // MCP callTool params
  {
    "toolName": "superglue_build_new_tool",
    "inputs": {
      "instruction": "Fetch user data from system A and send it to system B.",
      "systems": [
        { 
          "id": "systemA", 
          "urlHost": "https://api.systema.com",
          "credentials": { "apiKey": "system-a-key" }
        },
        { 
          "id": "systemB", 
          "urlHost": "https://api.systemb.com",
          "credentials": { "token": "system-b-token" }
        }
      ]
    }
  }

​

3. superglue_get_integration_code

• Description: Generate integration code for a specific tool. Use this to show users how to implement a tool in their applications.
• Input Schema: GenerateCodeInputSchema
  • toolId: The ID of the tool to generate code for.
  • language: Programming language for the generated code (typescript, python, or go).

• Example Usage (Conceptual MCP Call):

  // MCP callTool params
  {
    "toolName": "superglue_get_integration_code",
    "inputs": {
      "toolId": "tool-id-123",
      "language": "typescript"
    }
  }

​

4. superglue_run_instruction

• Description: Execute an instruction once without saving it as a persistent tool. Use for ad-hoc tasks that don’t need to be reused.
• Input Schema: RunInstructionInputSchema
  • instruction: Natural language instruction for the one-time execution.
  • payload: (Optional) Example JSON payload for the execution. This should be data needed to fulfill the request (e.g. a list of ids to loop over), not settings or filters.
  • systems: Array of SystemInputSchema defining the systems the execution can interact with.
    • id: Unique identifier for the system.
    • urlHost: Base URL/hostname for the system.
    • urlPath: (Optional) Base path for API calls.
    • documentationUrl: (Optional) URL to API documentation.
    • credentials: (Optional) Credentials for accessing the system.
  • responseSchema: (Optional) JSONSchema for the expected response structure.

• Example Usage (Conceptual MCP Call):

  // MCP callTool params
  {
    "toolName": "superglue_run_instruction",
    "inputs": {
      "instruction": "Get all active users from system A and create a summary report",
      "payload": { "includeMetrics": true },
      "systems": [
        { 
          "id": "systemA", 
          "urlHost": "https://api.systema.com",
          "credentials": { "apiKey": "system-a-key" }
        }
      ],
      "responseSchema": {
        "type": "object",
        "properties": {
          "totalUsers": { "type": "number" },
          "summary": { "type": "string" }
        }
      }
    }
  }

​

Dynamic Tool Execution

In addition to the static tools above, the MCP server dynamically creates execution tools for each of your existing Superglue workflows. These are named execute_{tool_id} and provide direct access to execute existing workflows.

​

Dynamic Tool Schema

Each dynamic tool has the following input schema:

• payload: (Optional) JSON payload data for the tool
• credentials: (Optional) Authentication credentials for the tool
• options: (Optional) Request configuration (caching, timeouts, retries, etc.)

The exact schema is derived from the tool’s inputSchema if available, otherwise it falls back to the flexible schema above.

​

Example Dynamic Tool Usage

// MCP callTool params for a tool with ID "stripe-to-hubspot-sync"
{
  "toolName": "execute_stripe-to-hubspot-sync",
  "inputs": {
    "payload": { "customerId": "cus_123" },
    "credentials": { 
      "stripeKey": "sk_test_...",
      "hubspotToken": "pat-na1-..."
    }
  }
}

​

Agent Workflow

The recommended workflow for agents using the Superglue MCP server:

1. DISCOVER: List available tools using the dynamic execute_{tool_id} tools or by calling superglue_execute_tool with known IDs
2. EXECUTE: Use superglue_execute_tool for existing tools, superglue_build_new_tool for new integrations, OR superglue_run_instruction for one-time ad-hoc tasks
3. INTEGRATE: Use superglue_get_integration_code to show users how to implement tools in their applications