Read-only SAP Datasphere MCP server — API Policy v4/2026 aligned, sibling to SAPBDCMCP.
The server demonstrates solid foundational quality with consistent tool naming, parameter descriptions, and proper schema definitions. All 8 tools follow verb_noun naming convention (list_, preview_, describe_, profile_, query_, search_, space_summary). Tool descriptions range from 47-98 characters, meeting the 10-1024 baseline. All parameters have type definitions and descriptions. However, the server lacks output schema documentation for any tool, making it impossible for LLMs to know what fields to expect in responses. Error handling descriptions are absent, no guidance on retryability, user-fixable vs fatal conditions, or recovery paths. The server is purely read-only (all tools marked READ_ONLY), which simplifies some risk patterns but means no destructive operations require confirmation flows. Tool descriptions are functional but could be more LLM-optimized with dependency hints and when-to-call guidance.
Describe the schema of a Datasphere asset with inferred data types and column statistics
List catalog assets for a given space
List all spaces in SAP Datasphere catalog
Preview data from a Datasphere asset with optional filtering and row limits
Analyze and profile a single column from a Datasphere asset
Execute a richer query against a Datasphere asset with filtering, ordering, and column selection
Search for assets across spaces in SAP Datasphere catalog
No output schemas documented for any of the 8 tools. LLMs cannot infer what fields to expect in responses, forcing guesswork about downstream tool chaining and field availability.
No error handling guidance provided. Tools lack descriptions of error conditions, retryability, recovery paths, or actionable error messages. This violates the recovery-guide pattern.
Missing pagination guidance for list_* and search tools. No documentation of total counts, next_cursor, or result limits. Large result sets could exhaust context windows.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-22 | D | 52 | 2026-07-28+ | v2 |
| 2026-03-09 | F | 36 | - | v1 |
Get a summary of a space including asset count breakdown by type and sample assets
Tool descriptions lack 'when to call' and dependency hints. E.g., preview_asset should hint that list_assets must be called first to discover asset names. This reduces discoverability for agents.
Parameters like 'space_id' and 'asset_name' accept only system identifiers, requiring agents to first call list_spaces and list_assets rather than using natural language names. Consider accepting human-friendly identifiers as fallback.