Introduction to MCP Server in SnapLogic

What is an MCP Server?

The Model Context Protocol (MCP) is a standardized way for AI applications (like Claude, custom AI agents, or chat interfaces) to connect with external tools, datasets, and APIs.

In SnapLogic, an MCP Server allows you to expose your custom SnapLogic pipelines as AI-ready tools. This means you can build a pipeline that performs a specific business task—such as querying a database, fetching real-time weather data, or updating an inventory system—and seamlessly make that capability available to an AI assistant.

Core Concepts

To successfully build and deploy an MCP Server in SnapLogic, you will work with two primary types of pipelines:

• The Tool Pipeline: The pipeline that actually performs the work (for example, "Get Weather" or "Lookup Customer"). This pipeline must be tagged as a Tool in pipeline properties configurations.
• The MCP Server Pipeline: The MCP server pipeline acts as an orchestrator that exposes your available tools and channels incoming requests to the correct Tool Pipeline. This pipeline must be tagged as an MCP server in pipeline properties configurations.

Quick Start

Ready to get your first SnapLogic MCP Server up and running? This section provides a high-level overview of the entire process.

Audience

The steps in the Quick Start guide assume you are familiar with the SnapLogic platform, its features and interfaces, and components.

Prerequisites

• Before you begin, ensure your SnapLogic organization is subscribed to the MCP Server feature and the MCP Server Snap Pack. If you do not see MCP-related Snaps or dialog options in your environment, contact your SnapLogic CSM to enable this functionality.
• Your SnapLogic environment must have a load balancer configured for your Snaplex. See MCP Server Setup for infrastructure requirements.

Quick Start Workflow

1. Step 1: Creating the tool pipeline
2. Step 2: Creating the main MCP Server Pipeline
3. Step 3: Registering your MCP Server
4. Step 4: Connect and Monitor your MCP Server

Creating the tool pipeline

Creating an MCP Server involves a two-layer architecture: Tool Pipelines (the workers) and the Main MCP Server Pipeline (the router).

The Tool Pipeline is where the actual business logic lives. For example, if you want your AI Agent to check the weather, this pipeline makes the API call to the weather service. The tool pipeline must have exactly one open input view and one open output view.

1. Build the Logic
   1. Download this pipeline and save it to your local machine.
   2. Open SnapLogic Designer.

   3. Click the Import icon and select TOOL_weather_forecast.slp from your local machine.

      

   4. Optional. To verify the pipeline executes correctly before using it as a tool, connect the disabled JSON Generator Snap to the HTTP Client Snap, then open the JSON Generator Settings and in the Snap execution field, select Execute and Validate.
   5. Optional. Save the Snap settings, and click to validate the pipeline. Skip these optional steps if you prefer to validate after completing the full Quick Start.
2. Tagging the pipeline as a Tool
   Note: The downloaded pipeline is already tagged as a Tool. This tells the MCP Server Pipeline that this pipeline is available as an AI tool. If you build your own tool pipeline from scratch, open the Edit Pipeline dialog and enable the Tool toggle.

   

3. Defining Parameters
   Note: The downloaded pipeline already has its parameters defined. The OpenAPI tab includes latitude and longitude as required number inputs, and the Info tab contains the description the MCP Server exposes to the AI. When building your own tool pipeline, define parameters in the Settings tab first, then complete the OpenAPI and Info tabs.

   

   

Creating the main MCP Server Pipeline

Now that your Tool Pipeline is ready, you need the MCP Server pipeline to manage incoming requests from the AI client.

Download a sample MCP Server pipeline to use as a starting point.

1. Import the pipeline
   1. Open SnapLogic Designer.
   2. Click the Import icon and select MCP_Server_Pipeline_As_Tool.slp from your local machine.
2. Tagging as an MCP Server
   Note: The downloaded pipeline is already tagged as an MCP server. SnapLogic automatically tags a pipeline as an MCP server when an MCP Server Router Snap is added to the canvas. No manual tagging is required.

   

3. Step A: Configure the Multi Pipeline Function Generator Snap
   Note: The downloaded pipeline already includes the Multi Pipeline Function Generator Snap (from the LLM Utilities Snap Pack). It reads the OpenAPI settings of your Tool Pipelines and exposes a formatted list of available tools. The Pipeline Path field is already set to reference the tool pipeline.

   

4. Step B: Connect the MCP Server Router
   Note: The downloaded pipeline already has all components connected. The MCP Server Router Snap requires no additional configuration beyond its connections. It routes tools/list requests by returning the tool list from the Function Generator, and routes tools/call requests to a Pipeline Execute Snap configured with the expression $sl_tool_metadata.path to dynamically call the matching tool pipeline. A Union Snap and Mapper Snap handle the combined response.

   

Registering your MCP Server

With your main pipeline open in the Designer, you can now register the server.

1. Open the Configure MCP Server dialog: In the SnapLogic Designer, with your Main MCP Server Pipeline open, locate and click the MCP Server icon in the top navigation ribbon.
2. Name and describe the purpose of this MCP Server:
   • Name: Enter a clear, identifiable name (for example, demo_inventory_mcp_server).
   • Description: Optional. Add context for human transparency (for example, "Provides AI access to inventory and weather systems").
3. Configure execution details:
   • Pipeline source: This will automatically select your currently open pipeline, but you can change it if necessary.
   • Snaplex: Select the execution environment where this server will run. Ensure this Snaplex has network access to the target systems your tools interact with.

     

   • Run Policy: Select one of the following:
     • Triggered (default)
     • Ultra
   • Optional. Modify timeout.
4. Authentication Policy: Select an existing policy from the dropdown, or click Create New Policy to create one directly from this drawer.
   Note: The following steps create a basic policy using the Anonymous Authenticator, suitable for development and testing. For production, use OAuth2 or API Key authentication instead.
   1. Click Create New Policy. A visual policy pipeline opens with stages: Validate, Authenticate, Authorize, Shape, and Transform.
   2. Configure Authentication:
      • Click the + under the Authenticate stage.
      • For development and testing, select the Anonymous Authenticator. This assigns a default role (for example, anonymous) to the incoming request.
      • For production environments, choose the MCP OAuth2 Client Credential or JWT Validator rules instead.
   3. Configure Authorization:
      • Click the + under the Authorize stage.
      • Select Authorize by role.
      • Configure it to allow access to the role assigned in the previous step (for example, anonymous). This ensures that only authenticated users with the correct role can trigger your MCP Server.
   4. Name your policy (for example, MCP_Basic_Auth_Policy) and click Save.

   

5. Review the form and click Save.

Upon saving, SnapLogic registers the server and generates a unique MCP Server URL.

Connect and Monitor your MCP Server

claude mcp add --transport http sl_mcp https://macawv2-qa-mcptest-fm.snaplogicdev.io/mcpserver/69af3e5b62747c47af2b6635/mcp
         Added HTTP MCP server sl_mcp with URL: https://macawv2-qa-mcptest-fm.snaplogicdev.io/mcpserver/69af3e5b62747c47af2b6635/mcp to local config
         File modified: /Users/admin101/.claude.json [project: /Users/admin1010]
/mcp  
 ⎿  MCP dialog dismissed                                                                                                               
 ❯ what is the weather in san francisco?                                                                                                 
         
Thinking…

  The user wants to know the weather in San Francisco. I have a weather forecast tool available. San Francisco's coordinates are
  approximately 37.7749° N, 122.4194° W.

⏺ sl_mcp - TOOL_weather_forecast (MCP)(latitude: 37.7749, longitude: -122.4194)
  ⎿  {
       "periods": "Name: This Afternoon\nTemperature: 87F\nWind: 7 mph NNW\nDetailedForecast: Sunny. High near 87, with
     temperatures falling to around 85 in the afternoon. North northwest wind around 7 mph.\n---\nName: Tonight\nTemperature:
     58F\nWind: 2 to 6 mph W\nDetailedForecast: Clear, with a low around 58. West wind 2 to 6 mph.\n---\nName:
     Wednesday\nTemperature: 86F\nWind: 2 to 6 mph NW\nDetailedForecast: Sunny. High near 86, with temperatures falling to around
     82 in the afternoon. Northwest wind 2 to 6 mph.\n---\nName: Wednesday Night\nTemperature: 58F\nWind: 2 to 6 mph
     W\nDetailedForecast: Clear, with a low around 58. West wind 2 to 6 mph.\n---\nName: Thursday\nTemperature: 87F\nWind: 2 to 7
     mph N\nDetailedForecast: Sunny, with a high near 87. North wind 2 to 7 mph."