---
sourceDocument: Brazil Customer Service Management
sourceDocumentLink: https://www.servicenow.com/docs/r/customer-service-management

 Release :

    - brazil

ft:locale :

    - en-US

ft:publication_title :

    - Brazil Customer Service Management

ft:clusterId :

    - csm

bundleId :

    - csm

workflow :

    - Customer and Industry


---

# CSM MCP Server

# CSM MCP Server {#ariaid-title1}

Release version: Brazil  
Updated September 10, 2026  
![](https://www.servicenow.com/docs/portal-asset/ico-clock) 4 minutes to read
Summarize  
![AI sparkle icon](https://servicenow.com/docs/portal-asset/ai-sparkle-icon) Summarized using AI  
This content was generated using new OpenAI-powered functionality. Results are provided on an as is basis and are not guaranteed to be accurate or complete.  

## Summary of CSM MCP Server

The CSM MCP Server acts as a secure intermediary that connects AI-enabled Model Context Protocol (MCP) client applications with your ServiceNow Customer Service Management (CSM) environment.
It enables customer service agents and managers to utilize case management and generative AI skill features through natural language conversations.
The server is active by default and registered asCSM MCP Server (sncsmmcpserver.csmdefault), supporting live data retrieval and action execution within ServiceNow.
Show full answer Show less  

## Key Features

* **Native Tool Exposure:** Allows external AI agents and large language model (LLM)-based tools to interact with ServiceNow capabilities via MCP.
* **Instant Onboarding:** Supports quick and automated authentication-compliant connections for generative or agentic AI platforms.
* **Controlled Access:** Enforces all configured permissions, rules, and access controls regardless of the access channel.
* **Complete Observability:** Provides usage, latency, and connection health metrics through AI Control Tower for comprehensive monitoring.
* **Case Management Tools:** Supports subflows and actions like retrieving case lists, case details, case tasks, and task details.
* **Generative AI Skill Tools:** Includes features such as case summarization, draft resolution notes generation, sentiment analysis with access enforcement, and AI-drafted activity responses with multiple response modes.
* **Supported MCP Primitives:** Supports only the tools primitive for executing capabilities; resources and prompt primitives are intentionally not supported to mitigate prompt injection risks.
* **Supervised and Autonomous Execution:** Execution mode is controlled by the connecting MCP client, enabling organizations to apply governance policies at the client level.
* **Usage Tracking and Observability:** All MCP server traffic and telemetry data are accessible in AI Control Tower depending on environment setup.
* **MCP Apps Support:** Supports structured text and JSON responses, with native support for custom MCP Apps (HTML/UI) in the latest release, subject to administrator configuration.

## User Roles and Permissions

To operate the CSM MCP Server, users require specific roles:

* **admin:** Activates the CSM MCP Server.
* **sncustomerserviceagent:** Reads customer case records, work notes, and generates resolution notes.
* **sncustomerservice.consumeragent:** Reads consumer case records, work notes, and generates resolution notes.
* **sncsmmcpinvoke:** Required to invoke key case management tools like *getcustomerservicecases* and *getcasetasks*.

## How It Works

The CSM MCP Server operates by receiving requests from MCP client applications (e.g., Moveworks, Claude) where customer service agents ask questions or request actions. It authenticates users securely using OAuth 2.1 and JWT with mutual TLS transport, enforces role-based access control, and then retrieves or updates data in the ServiceNow instance. The server then returns actionable results to the AI client application, enabling seamless interaction with live CSM data and AI-driven assistance.  
The CSM MCP Server connects AI-enabled Model Context Protocol (MCP) client applications to your ServiceNow environment, enabling case management and generative AI skill features for customer service agents and
managers.

The CSM MCP Server operates as a secure intermediary between any MCP client application and your ServiceNow instance. The server is registered as CSM MCP Server (`sn_csm_mcp_server.csm_default`) and is active by default. Customer service agents use subflows, actions, or AI skills through an MCP client such as Moveworks or Claude. The CSM MCP Server retrieves data live from the ServiceNow Customer Service Management environment and performs actions through natural language conversation.{#csm-mcp-server__csm-mcp-server-para-1}

## Key benefits {#csm-mcp-server__csm-mcp-server-benefits}

The CSM MCP Server provides the following benefits:{#csm-mcp-server__csm-mcp-server-para-2}

Native tool exposure
:   External AI agents, LLM-based tools (CLI, chat, etc) can connect with and use ServiceNow capabilities as tools through MCP.

Instant onboarding
:   Any generative or agentic AI platform can be connected in minutes with automated, authentication-compliant setup.

Controlled access
:   All permissions, rules, and access controls that you set up are enforced, regardless of access channel.

Complete observability and auditability
:   Usage, latency, and connection health across every server, tool, and client are available through AI Control Tower.

## CSM MCP Server capabilities {#csm-mcp-server__csm-mcp-server-capabilities}

The CSM MCP Server currently exposes 6 tools: 2 case management tools and 4 generative AI skill tools. This is within the recommended range of 20-30 tools per MCP server for optimal performance and
manageability.{#csm-mcp-server__csm-mcp-server-para-3}

Each tool carries platform-level annotations indicating the tool is read-only and whether the tool is available to all users or restricted by role.{#csm-mcp-server__csm-mcp-server-para-4}

## Case management tools {#csm-mcp-server__csm-mcp-server-case-mgmt}

The CSM MCP Server handles these core subflows and actions for case management:{#csm-mcp-server__csm-mcp-server-para-5}

* Gets a list of cases for given criteria or gets details of a specific case
* Gets a list of case tasks for given criteria or details of specific case tasks
* Retrieves case and case task details based on the case number
{#csm-mcp-server__csm-mcp-server-list-1}

## AI skill tools {#csm-mcp-server__csm-mcp-server-genai}

The CSM MCP Server handles these AI skill tools:{#csm-mcp-server__csm-mcp-server-para-6}

* Summarizes case activity, key actions, and current status
* Generates a draft resolution notes entry for the case
* Evaluates the sentiment of a CSM case using the Sentiment Analysis generative AI skill. Returns sentiment score, trend, and reasoning. Enforces record-level read access before invoking the skill.
* Generates an AI-drafted activity response for a CSM case using the Activity Response Generation skill. Supports three response modes: Respond, Follow-Up, Summarize.
{#csm-mcp-server__csm-mcp-server-list-2}

## Supported MCP primitives {#csm-mcp-server__csm-mcp-server-primitives}

The CSM MCP Server supports the tools primitive only. This MCP server does not support resources or prompt primitives. This means:

* Resources unavailable: Claude can't retrieve live ServiceNow data (such as records, Snowflake analytics, or Seismic assets) to pass into MCP tool calls. All tool execution must rely on data provided explicitly in the conversation.
* Prompt primitives unavailable: Claude can't use prebuilt or dynamic prompt templates to adapt its reasoning based on context. Tool behavior remains consistent across all invocations, regardless of the task or situation.
{#csm-mcp-server__ul_x25_cdc_lkc}The resources and prompts primitives aren't supported. This is a deliberate design decision driven by prompt injection attack risk mitigation.

## Supervised and autonomous execution {#csm-mcp-server__csm-mcp-server-execution}

Whether a tool call runs in supervised or autonomous mode is a setting controlled by the connecting MCP client, not a setting available at the MCP server level. For example, this setting is configured in
Claude desktop when you add a third-party MCP server (like GitHub or Slack) as a connector. Your organization's governance policies regarding tool execution mode should be configured at the client level.{#csm-mcp-server__csm-mcp-server-para-9}

## Usage tracking and observability {#csm-mcp-server__csm-mcp-server-observability}

MCP server traffic, including billing, usage metrics, and debugging telemetry, is available in the AI Control Tower. CSM MCP Server usage visibility depends on your ServiceNow environment configuration and AI Control Tower setup.{#csm-mcp-server__csm-mcp-server-para-10}

## MCP Apps support {#csm-mcp-server__csm-mcp-server-apps}

The CSM MCP Server tools return structured text and JSON responses. The platform natively supports MCP Apps (custom HTML/UI) in the latest release, allowing custom interfaces to be rendered by the client with
configurable permissions (microphone, camera, geolocation, clipboard). Check with your ServiceNow administrator regarding whether CSM ships custom MCP App UI or relies on text/JSON-only responses.{#csm-mcp-server__csm-mcp-server-para-13}

## CSM MCP Server users {#csm-mcp-server__csm-mcp-server-users}

The following roles are required for CSM MCP Server users:{#csm-mcp-server__csm-mcp-server-para-14}
{#csm-mcp-server__csm-mcp-server-table-1__entry__2}

| Role | Description |
|-|-|
| admin | Activates the CSM MCP Server |
| sn_customerservice_agent | Reads CSM customer case records, work notes, and generates resolution notes for customer cases |
| sn_customerservice.consumer_agent | Reads CSM consumer case records, work notes, and generates resolution notes for consumer cases |
| sn_csm_mcp_invoke | Required to invoke get_customer_service_cases tool and get_case_tasks tool |
| sn_csm_mcp_invoke | Required to invoke get_case_tasks tool |
[Table 1. CSM MCP Server user roles]

{#csm-mcp-server__csm-mcp-server-table-1}

## Server operation {#csm-mcp-server__csm-mcp-server-operation}

The CSM MCP Server operates through the following process:{#csm-mcp-server__csm-mcp-server-para-8}

1. A Customer Service Management MCP Server user asks a question using an MCP client. For example, a live agent opens their MCP client application such as Moveworks or Claude. The agent can ask for a list of open cases by priority or to summarize a case.
2. The MCP client application sends the question to the CSM MCP Server using the Model Context Protocol.
3. The CSM MCP Server authenticates the user against the ServiceNow role-based access control. Authentication uses OAuth 2.1 and JWT with mTLS transport for secure communication.
4. The CSM MCP Server retrieves the requested data or performs the requested action in the ServiceNow instance.
5. The CSM MCP Server returns the result to the AI client application, which displays the result to the user.
{#csm-mcp-server__csm-mcp-server-list-3}