🧾 MCP Server for SS&C Next Generation

This project provides an MCP-compatible server that connects AI agents to SS&C Next Generation, enabling automated execution of business processes via the REST API.

Designed with flexibility in mind, this server allows you to:

Easily define and manage tools using a simple CSV configuration, without modifying source code.

Automatically submit items to Work Queues, a key component of Next Gen RPA orchestration.

Trigger automation flows in response to new queue items, creating a seamless bridge between AI-driven decisions and RPA actions.

Whether you're integrating conversational agents, scheduling systems, or custom applications, this server gives you a low-code interface to dynamically register actions and automate task submission with minimal overhead.

🚀 Installation & Setup

This section helps you set up and run an MCP-compatible server that interacts with SS&C Next Gen REST API.

🔧 Prerequisites

• Python 3.10 or later
• uv (recommended) package manager
• MCP-compatible client (e.g., Claude Desktop)
• SS&C Next Generation Environment(Japanese/English)

---

📥 Installation Steps

1. Prepare Next Generation Service Account

Refer to the link below to create a service account.

Service Account

2. Prepare Next Generation Work Queue

Work queue preparation is required when using dynamic tools. Please refer to the following link to create a work queue.

Work Queues

When combining with automated processing, it is necessary to check the following Automation Flow information and link it to the work queue

Automation Flow

3. Prepare the project directory

git clone https://github.com/abe1bp/mcp-server-nextgen.git
cd mcp-server-nextgen

---

⚙️ Add MCP Server to claude_desktop_config.json

To connect Claude Desktop to your server, add:

{
  "mcpServers": {
    "mcp-server-nextgen": {
      "command": "uv",
      "args": [
        "--directory",
        "{folder path}",
        "run",
        "mcp-server-nextgen"
      ],
      "env": {
        "OAUTH_CLIENT_ID": "{your-client-id}",
        "OAUTH_CLIENT_SECRET": "{your-client-secret}",
        "OAUTH_TOKEN_URL": "https://{tenant-domain}/realms/{tenant-id}/protocol/openid-connect/token",
        "BASE_URL": "https://{tenant-domain}/regions/{region}/api/rpa/rest/v1"
      }
    }
  }
}

{folder path} is the folder prepared in Prepare the project directory.

Cconfigure the {your-client-id} and {your-client-secret} of the Service Account created in Step 1 of the Installation Steps.

For each URL parameters, please refer to the links provided below. Next Generation REST API Then restart Claude Desktop and select mcp-server-nextgen.

---

📌 Purpose

This Python-based MCP server leverages the Model Context Protocol (MCP) to:

• Connect to Next Generation REST API
• Authenticate via OAuth2 and expose secure MCP endpoints
• Dynamically define tools to submit items to Next Generation Work Queues

---

📁 Key Components

🛠 Tool Types

🧩 Retrive Tools (predefined)

• retrieve-automation-flow-list
• retrieve-activity-log-list
• retrieve-digital-worker-list
• retrieve-session-list

🧩 Start automation flow Tool (predefined)

• start-automation-flow

---

📄 Dynamic Tools (CSV-defined)

CSV Configuration for Dynamic Tool Loading

The workqueues.csv file defines tools that are dynamically loaded at server startup. Each row in the CSV represents a tool that can submit an item to a specific work queue.

Column Descriptions

Here is a clearer breakdown of the required columns:

Column	Required	Description
workqueueid	Yes	The unique identifier (UUID format) of the work queue where the item will be submitted.
name	Yes	A short, unique identifier for the tool. Used as its registered name.
description	Yes	A brief explanation of what the tool does. Displayed to clients and used to assist tool discovery.
inputSchema	Yes	A JSON Schema (as a string) defining the input structure required by the tool. Each input parameter must include a description to be usable by the client.
keyValue	No	A key or identifier that will be passed to the queue item (e.g., an Order ID). Can be a fixed value or a reference.
priority	No	An integer indicating the priority of the submitted item (e.g., 0 = lowest, 100 = highest).
status	No	Initial status assigned to the item when submitted (e.g., New, Pending).
tags	No	Comma-separated tags to help categorize or filter items in the queue.

Example

workqueueid,name,description,inputSchema,keyValue,priority,status,tags
aaaaaaaa-bbbb-cccc-dddd-eeeeeeeeeeee,submit-order,"Submit an order",{"type":"object","properties":{"OrderId":{"type":"string","description":"Order ID"}}},"Order001",50,"New","order,urgent"

Notes

• Make sure inputSchema is a valid JSON object serialized as a string.
• The description fields within inputSchema are essential for UI rendering and interaction.
• You can omit optional columns if not needed, but headers must still be present.

This format allows you to configure and register dynamic tools for queue submissions without code changes.

📌 Dynamic tools are flexible:

• You can define multiple tools per work queue.
• Set proper name, description, inputSchema, and parameters to trigger item submission only when needed.

---

📤 Prompts

This is used when Client id and Client Secret are not set as environment variables.

Prompt Name	Purpose
Set-NextGen-Login	Set Client ID / Secret manually

---

🧪 Example: Tool Execution

✅ Call retrieve-digital-worker-list

Returns the list of digital workers and updates MCP clients via resources_changed.

✅ Call submit-order (CSV-defined tool)

Posts an item to a work queue:

{
  "keyValue": "Order001",
  "priority": 50,
  "status": "New",
  "tags": ["order", "urgent"],
  "data": "<base64-encoded XML>"
}

---