ServiceNow MCP data reads: API Access Volume and Assist metering clarification
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
yesterday
Hi community,
Could someone clarify the end-to-end consumption rules for read-only data access through ServiceNow’s native MCP Server Console?
Example scenario
ChatGPT → ServiceNow native MCP Server → registered Scripted REST API tool → GlideRecord reads from the incident table on the same instance → JSON results returned to ChatGPT through MCP.
Assume the script only retrieves records. It does not invoke a Now Assist/GenAI skill, an AI Agent, a subflow, or additional integrations with external systems. This scenario uses ServiceNow’s native MCP Server Console, rather than an externally hosted custom MCP server wrapping the Table API.
Main question
Does this complete operation consume API Access Volume in addition to Assists?
I would like to distinguish consumption at the MCP endpoint from consumption associated with the backing Scripted REST API, and understand the total consumption for the complete data-retrieval operation.
What I found in the public documentation
The ServiceNow Otto Overview, effective November 12, 2026, lists:
- MCP server call: 1 Assist per call, on page 14.
- Action Fabric record reads: 1 Assist per 100 records read through ServiceNow MCP Server, on page 4, excluding reads performed as part of a GenAI skill or agentic workflow.
The Data Fabric Credit Overview separately describes API Access Volume as data egress through ServiceNow public APIs. This is a data-volume measure, distinct from API request counts.
Questions
- Is the entire example operation excluded from API Access Volume, or only the MCP endpoint’s traffic? Does executing the registered Scripted REST API create separately metered data egress?
- If API Access Volume applies, which response is measured: the backing REST API response, the MCP response, or another boundary?
- For this scenario, how do the MCP-call and record-read Assist entries interact? Are they additive, or does an applicability rule determine which entry is charged?
- Do these rules depend on the entitlement/SKU, platform release, or MCP Server Console version?
Which dashboards or usage records should be compared to validate this—particularly Assist consumption versus the Data Egress tab in Inbound API Integration Usage?
Links to official documentation or a relevant KB explaining the metering boundaries and exclusions would be especially helpful. I understand that account-specific contractual applicability needs confirmation with the account team; I am seeking clarification of the documented metering behavior.
This is a design question, not a report of measured results from an instance.
Thank you!