What is MCP Server Console?
MCP Server Console is ServiceNow's purpose-built control plane for publishing, governing, and observing AI agent connections. It gives any standards-compliant MCP client governed access to ServiceNow capabilities, from Now Assist skills to platform workflows and AI Agents, without custom integrations or one-off connectors.
When we introduced MCP Server Console in December 2025, the promise was straightforward: give any AI agent a governed, standards-compliant path into ServiceNow, no custom integration, no one-off connectors. The launch established the foundation, shipping Now Assist Skills as MCP tools and a Quickstart Server ready to use out of the box.
Since then, we have expanded the tool surface significantly. This article covers the full journey, what shipped at launch, what's new in v1.4, and how to set your servers up for success. If you are coming to MCP Server Console for the first time, we recommend starting with Bring the ServiceNow AI Platform to Any Employee Experience for an overview of the platform and the MCP protocol, watch the AI Academy setup video for initial server configuration, or consult the product documentation before returning here.
In this article:
✓ What shipped at launch: v1.2 (Q4 2025)
✓ What's new in v1.4: Knowledge Graph, Subflows, Scripted REST APIs, and AI Agent support
✓ What your agents can do today
✓ Best practices for setup
✓ Governance, observability, and technical requirements
🚀 What Shipped at Launch: v1.2, Q4 2025
The v1.2 release established the console itself and the first tool type, giving administrators a single place to publish, curate, and govern ServiceNow tools for any MCP client.
|
1
MCP Server Console A purpose-built control plane for every AI agent connection. Create multiple MCP servers, each scoped to a specific use case or persona, and manage a complete tool catalog from one place. Activate, deactivate, and curate exactly which tools each client can access, with full server lifecycle management built in. |
2
Now Assist Skills as Tools Pre-built and custom Now Assist skills packaged as governed MCP tools, discoverable and callable by any standards-compliant MCP client. The Quickstart Server ships pre-configured with incident summarization and case summarization, no setup required to get started. Custom skills built with Now Assist Skill Kit are supported alongside ServiceNow-published capabilities. |
MCP Server Console — Quickstart Server with incident and look up incident records tools configured
🆕 What's New in v1.4
v1.4 expands the tool surface from skills to the full ServiceNow AI Platform, giving any authenticated MCP client the ability to reason about relationships, trigger workflows, reach custom data, and invoke AI Agents directly.
Want to see these configured end to end? Watch the AI Academy walkthrough below, then follow the written guidance for each tool type.
🤖 What Your Agents Can Do Today
| What the user asks | Capability |
|---|---|
| "Summarize incident INC39582 for me" | Now Assist Skills |
| "How many open incidents does my team have, and which CIs are affected?" | v1.4 Knowledge Graph |
| "Create a PTO request for next week and notify my manager" | v1.4 Subflows and Actions |
| "Show me our team's SLA performance against our KPIs this quarter" | v1.4 Scripted REST APIs (Full CRUD) |
| "Run the change risk assessment agent on CHG0012045" | v1.4 Support for Agents via Scripted REST |
⚙️ Best Practices for Setup
|
1
|
Design servers around use cases, not tools The most effective MCP Server Console deployments create one server per use case, an IT server, an HR server, a CSM server, rather than one large server containing every available tool. This keeps tool catalogs focused, makes governance easier, and prevents the token bloat that occurs when an MCP client receives an oversized list of tools it will rarely use. AI clients load the full tool catalog at connection time. The larger that catalog, the more tokens consumed before a single query is answered. Keep servers lean and scoped. |
Our product teams are releasing pre-built domain-specific MCP servers. Currently we have the following:
Contact your account team for details and roadmap.
|
2
|
Write rich tool descriptions Tool descriptions are not documentation for humans, they are the instructions an AI agent uses to decide whether and how to call a tool. Every description should answer two questions: what does this tool do, and when should it be called? Example: Incident lookup tool description "Retrieves a ServiceNow incident record by incident number. Call this tool when the user references a specific incident by number and needs details such as state, priority, assignment group, or recent work notes. Do not call this tool for general incident searches or list queries." ServiceNow ships tools with rich descriptions by default. When creating custom tools, particularly Scripted REST tools, invest time in the description. The quality of the description directly determines how reliably your AI client will use the tool correctly. |
|
3
|
Provide explicit inputs for Scripted REST tools When calling a Scripted REST tool, the calling agent must supply all required inputs explicitly in the query. These tools do not infer context the way a conversational skill might. If a required parameter, such as a record sys_id, a date range, or a filter value, is not present in the request, the tool cannot complete the task. Structure your agent prompts and tool configurations to declare inputs clearly. |
|
4
|
Be deliberate about tool-to-server mapping A single tool can be added to multiple servers. This is useful for shared capabilities, but carries an important responsibility: any change to that tool affects every server it is mapped to. Before editing a tool's definition, description, or inputs, verify which servers it belongs to and assess the downstream impact. MCP Server Console shows you the servers a tool is associated with; consult that view before making changes. |
|
5
|
Match server scope to app scope Global tools must live on global servers. If a tool is defined within a scoped application, it must be added to a server within that same scope, it cannot be added to a global server. Mixing scopes causes tools to become unavailable at runtime. When planning your server architecture, align server scope to the applications that own the tools you intend to expose. |
🔒 Governance, Control, and Observability
Every tool type ships within the same governance framework. MCP Server Console provides built-in access control at the account, entitlement, and client levels. Administrators control which tools are exposed to which clients and manage OAuth client registrations through the console.
📋 Technical Requirements and Feature Summary
To activate all current tool types:
- Zurich Patch 9 or Australia Patch 2 minimum (ZP7/8 or AP1 for Skills as tools only)
- Now Assist for ITSM, HRSD, or CRM store app (latest version)
- MCP Server Console store app, available on the ServiceNow Store
- MCP client of your choice
- AI Control Tower with AI Gateway (optional, recommended for usage telemetry, governance, and observability)
| Feature | Available |
|---|---|
| MCP Server Console | Q4 2025 |
| Now Assist Skills as Tools | Q4 2025 |
| Knowledge Graph as Tools | v1.4 |
| Subflows and Actions as Tools | v1.4 |
| Scripted REST APIs (Full CRUD) as Tools | v1.4 |
| Support for AI Agents via Scripted REST | v1.4 |
🚀 Get Started
The MCP Server Console store app is available today on the ServiceNow Store. Additional resources:
- MCP Server Console product documentation
- AI Academy: Bring Now Assist to your AI Experience with ServiceNow's MCP Server (initial setup walkthrough)
- REST API Eligibility Criteria for MCP Tool Integration (KB2952534)
- Bring the ServiceNow AI Platform to Any Employee Experience (community overview)
