ServiceNow MCP Tools for AI Agents
Let an agent search incidents, read knowledge articles, create and update records, and order items from the ServiceNow catalog.
Available Tools
21 tools available in this integration
Servicenow.AddComment Add a comment or internal work note to a work record.
Servicenow.CountRecords Count work records matching a filter, optionally broken down by a field.
Servicenow.CreateCatalogRequest Order an item from the service catalog.
Servicenow.CreateComment Add a comment or internal work note to a work record.
Servicenow.CreateIncident Open a new incident.
Servicenow.GetCatalogItem Read a catalog item's details and the options required to order it.
Servicenow.GetConfigurationItem Read a configuration item together with what it depends on and what depends on it.
Servicenow.GetKnowledgeArticle Read a knowledge article's full body.
Servicenow.GetRecord Read a work record together with its full comment and work-note history.
Servicenow.GetRecordCounts Count work records matching a filter, optionally broken down by a field.
Servicenow.GetRecordFieldChoices Discover the valid choice values for a work-record field on this instance.
Servicenow.ListFieldChoices Discover the valid choice values for a work-record field on this instance.
Servicenow.OrderCatalogItem Order an item from the service catalog.
Servicenow.SearchAssignmentGroups Find ServiceNow assignment groups by full or partial name, to route work to them.
Servicenow.SearchCatalogItems Find service catalog items matching a described need.
Servicenow.SearchConfigurationItems Find configuration items (systems) by name, optionally narrowed to a CMDB class.
Servicenow.SearchKnowledgeArticles Search the knowledge base for articles matching a described problem or topic.
Servicenow.SearchRecords Find work records of one type by text, assignment, state, priority, or recency.
Servicenow.SearchUsers Find ServiceNow users by name or email, to assign or contact them.
Servicenow.UpdateRecord Change one or more fields on a work record.
Servicenow.WhoAmI Return the authenticated ServiceNow user's profile.
Why agent-optimized tools matter
Built for intent, not API schemas
Ask the agent to "find the high-priority incidents assigned to the network team and summarize their latest work notes." Arcade picks the right ServiceNow actions without guessing record types or record identifiers.
Up to 100x fewer tokens
ServiceNow searches return matching record details without loading full comment and work-note histories, so the agent can narrow the queue first and fetch the complete record context only when it needs it.
Complete the workflow end to end
Search an incident, read its history, find the right user or assignment group, update the record, and add a work note through one toolkit designed around the complete ITSM task.
Make actions predictable and controlled
Arcade tools return structured results and useful errors across users and runs. Teams can apply scope restrictions, redaction, and approval policies so agents follow the same controls every time.
Securing ServiceNow
The more capabilities agents get, the more useful they are. However, more capabilities bring more risks of undesired effects such as data loss and exfiltration.
This is why Arcade provides you with contextual rules that dictate how and when a tool is allowed to run. Arcade’s contextual access lets you govern tool visibility and behavior at every stage of execution and control who can see a tool, what inputs are allowed, and what comes back.
Learn about Contextual Access PoliciesFor ServiceNow, you can implement some of these policies:
- Hide Servicenow.OrderCatalogItem from agents unless the roles your IDP marks as approved requesters are active, so an agent cannot order equipment or paid services for your organization.
- Restrict Servicenow.UpdateRecord to agents operating under your incident-manager or change-manager roles, so routine automation cannot alter ownership, priority, state, or approval fields.
- Show Servicenow.CreateIncident only to agents serving your support or operations roles, so untrusted workflows cannot flood your incident queue with fabricated tickets.
- Hide Servicenow.CreateCatalogRequest from agents unless the roles your IDP marks as authorized requesters are active, so an agent cannot submit purchases or access requests for your organization.
- Restrict Servicenow.CreateComment to agents operating under your support or case-management roles, so general-purpose workflows cannot add notes to other people's work records.
- Require human approval before Servicenow.OrderCatalogItem runs, so an agent cannot place a service request or purchase without a person confirming the details.
- Block Servicenow.CreateIncident when the description contains credentials, access tokens, or customer secrets, so sensitive data is not copied into incident records.
- Refuse Servicenow.AddComment when the comment contains passwords, tokens, or payment details, so an agent cannot publish secrets in a work record's history.
- Require human approval before Servicenow.CreateCatalogRequest runs, so a person confirms the requested item, quantity, and fulfillment details before submission.
- Block Servicenow.CreateComment when the comment contains passwords, tokens, payment details, or unredacted customer data, so sensitive content cannot be added to a work record.
- Redact Servicenow.GetRecord results to remove phone numbers, personal email addresses, credentials, and full comment or work-note bodies before they reach the model.
- Strip personal email addresses, phone numbers, and other private contact fields from Servicenow.SearchUsers results, leaving only the identity details needed for assignment.
- Deny Servicenow.GetKnowledgeArticle output when the article is marked confidential, restricted, or under legal hold, so its contents never reach the model.
- Redact Servicenow.CountRecords breakdowns for groups containing fewer than five records, so the model cannot infer an individual's case volume from a small team.
- Strip internal-only labels and values from Servicenow.ListFieldChoices results before they reach the model, so agents cannot learn restricted workflow states or administrative routing codes.
How to use ServiceNow with your agent
Pick an MCP client below and follow the setup to connect ServiceNow to your agent. Each user authorizes their own ServiceNow account.
Your questions, answered
A ServiceNow MCP server exposes ServiceNow ITSM capabilities as structured actions an AI agent can call through the Model Context Protocol. After a user authorizes their account, the agent can search and read work records, consult knowledge articles, update incidents, and order catalog items within the user's permissions.
ServiceNow does not provide the first-party MCP endpoint used here. Arcade provides the MCP runtime with per-user OAuth, agent-optimized actions, approval workflows, redaction, and scope controls for connecting ServiceNow to different agents and clients.
Arcade agents can search, count, and read incidents and other work records; inspect comments and work notes; create incidents; add comments or internal notes; update record fields; find users and assignment groups; search configuration items; read knowledge articles; and order catalog items.
Each user authorizes their own ServiceNow account through OAuth. The resulting access is limited by the permissions and roles assigned to that ServiceNow user, including the rights needed to read records, create or update work, and order catalog items.
Yes. Each user authorizes their own ServiceNow account through OAuth, and actions run under that user's authorization and ServiceNow permissions. One user's records and available actions are not automatically available to another user.
Yes. Contextual Access policies can require human approval before consequential ServiceNow actions, including creating or updating records, adding work notes, and ordering catalog items. The call proceeds only after the policy's conditions have been satisfied.
Arcade proxies ServiceNow API calls and does not persist your ServiceNow content as part of normal tool execution. Credentials are brokered and stored encrypted at rest, and the model never receives them.
Any framework or client that can connect to Arcade's MCP gateway or tool APIs works, including LangChain, OpenAI Agents, CrewAI, Claude Desktop, Claude Code, Cursor, Visual Studio Code, Microsoft Copilot Studio, and GitHub Copilot.
Arcade refreshes access while the user's ServiceNow authorization remains valid. If it is revoked, rotated, or can no longer be refreshed, calls stop until the user reconnects their ServiceNow account.
ServiceNow instance and API limits still apply to calls made through Arcade, along with any applicable Arcade request limits. Exact quotas depend on the ServiceNow instance and its configuration.