AI & Analytics Legends The knowledge platform for SAP Analytics Sign inSign up
Guide

SAP Datasphere MCP server: community options and checks

SAP Datasphere MCP server: community options and checks — Analytics Legends section illustration for SAP analytics market intelligence (Datasphere, BDC, SAC, BW/4HANA)

As of 2026-10-18

Edited by , sole editor and publication director — 27 years in the SAP ecosystem. How these pages are produced, and where automation is used.

No official SAP MCP server for Datasphere was found on SAP Help on 18 October 2026. The option that exists is a community server (MIT, version 2.x, in active development) that explores metadata, runs analytic queries, extracts data and inspects lineage, with configurable PII masking. A separate community server covers Delta Sharing in Business Data Cloud.

What exists

SAP Datasphere MCP (community) lets an assistant explore Datasphere metadata, run analytic queries, extract data and inspect lineage, with configurable PII masking. It supports both stdio and HTTP, is MIT licensed, and the catalogue labels it active development (2.x). It is not an SAP product.

SAP Business Data Cloud MCP (community, MIT, stdio) is narrower: seven tools for Delta Sharing in BDC, covering publishing data products, validating shares and provisioning shares end to end. This one acts rather than reads, so it deserves more caution.

Official: none verified. The Integration Suite MCP Gateway can sit in front of APIs and external MCP servers, but that is a general control plane, not a Datasphere server.

What an assistant can do

With the Datasphere server: discover which spaces, views and tables exist, understand their columns, query modelled data, and follow lineage to see where a figure comes from. Lineage is the most valuable of these for a consultant doing impact analysis or an audit trail. The extraction capability is the one to watch, because it moves data out of Datasphere into an assistant's context.

What it cannot give you is the business meaning. Descriptions, semantic types and business names only help if they are maintained in the model; a poorly documented space will produce a confident but poorly informed assistant. Catalogue hygiene decides the quality of the answers.

A reasonable first use is a metadata-only pilot: the assistant lists spaces and views, reads descriptions and traces lineage, and never touches rows. That alone supports documentation, impact analysis before a change, and onboarding of new team members. Add querying later, on a dedicated space, once masking and identity have been tested. Extraction comes last, if at all.

Local or remote

The Datasphere server supports both transports; the BDC one is stdio only. A stdio setup runs with the credentials of whoever starts it. An HTTP deployment is a service you host, secure and monitor yourself, and the catalogue does not describe the authentication of either server, so do not assume OAuth or per-user passthrough.

Checks before production

Identity: does the server query as one technical user, or as the individual? A single technical user bypasses the row-level security you built in the Datasphere security model. Scope: expose only a space or view set designed for AI use, and give the account read access only. Masking: PII masking is configurable, so confirm it is switched on and test it on a real sensitive column rather than trusting the default. Extraction limits: cap rows returned. Write tools: for the BDC server, require explicit approval before provisioning a share. Audit: keep a log that ties a query to a person. And remember that any data returned goes to the model provider: check that against your data-processing terms.

Joule, BDC agents and agent platforms

SAP's own route for AI over BDC data is Joule and its agents, which stay inside SAP's governance. A community MCP server is the route for external assistants such as Claude. The design choice is between the two: SAP's embedded assistants give you governance but only SAP's tooling; an MCP server gives openness but you carry the governance. Where Joule Studio can attach MCP servers to an agent, the same caution applies.

Analytics Legends publishes its own read-only MCP server for the consulting market around Datasphere: firms, day-rate benchmarks, opportunities and news. Its public tranche needs no key.

What we cannot assert

Authentication of both community servers is not stated in the sources we verified (18 October 2026).

Frequently asked

Is there an official SAP Datasphere MCP server?

Not that we could verify; a SAP Help Portal search on 18 October 2026 found none. The available server is community-built.

Can it mask personal data?

The community server offers configurable PII masking. Test it on your own sensitive columns.

What about Business Data Cloud?

A separate community server covers Delta Sharing with seven tools for publishing data products and provisioning shares.

Can I put it behind a gateway?

Yes in principle: SAP Integration Suite MCP Gateway fronts external MCP servers with authentication, rate limiting and monitoring, and needs the Premium or Enhanced edition and Integration Cell.

Engage the editor

This guide is written by . 27 years of SAP — Datasphere, Business Data Cloud, SAC Planning, Joule. Freelance or permanent · Hybrid / remote · Anywhere in EMEA. Available from 19 October 2026.

Profile and contact

What this page is built on

External sources

Read next

Guides that answer with this page

These guides cite this page as one of the sources their answer rests on.