mdx-mcp
Enables natural language querying of OLAP/SSAS cubes with verified MDX generation, self-consistency checks, and honest abstention when queries are ambiguous.
README
mdx-mcp
An open-source, verified natural-language → MDX MCP server for OLAP cubes.
Ask an OLAP/SSAS multidimensional cube questions in plain English and get back a verified answer — the number, the MDX that produced it, and an honest abstain/clarify when the query is ambiguous. Cross-platform (XMLA), provider-agnostic (Claude by default), and MCP-native.
question ──▶ introspect the cube ──▶ generate k candidate MDX ──▶ execute (read-only)
──▶ self-consistency verify ──▶ { status, value, mdx, agreement }
Why it's different
- Verified, not vibes. It runs k diverse candidate queries and only answers when they agree; on divergence it abstains or asks to clarify instead of guessing a number.
- Works on any cube. Cube-schema introspection auto-grounds generation — no per-cube hand-authoring.
- Read-only by construction. Only
SELECT/WITHMDX ever executes. - Open-core. Three clean seams (
LLMClient,MdxExecutor,Verifier) let you plug in a private trust layer without forking. SeeARCHITECTURE.md.
Install
pip install mdx-mcp[claude] # engine + Claude reference provider
The XMLA executor and cube introspection use only the Python standard library.
Configure (env)
export MDX_MCP_ENDPOINT="http://your-ssas-host/olap/msmdpump.dll" # XMLA endpoint
export MDX_MCP_CATALOG="YourDatabase"
export MDX_MCP_CUBE="Your Cube"
export MDX_MCP_USER="user" # optional (basic auth)
export MDX_MCP_PASSWORD_FILE="/run/secrets/olap_pw" # or MDX_MCP_PASSWORD
export ANTHROPIC_API_KEY="sk-..." # for the default Claude producer
Run
mdx-mcp # starts the MCP server (stdio)
Tools
| Tool | Purpose |
|---|---|
mdx_introspect |
the cube's grounding block (measures / dimensions / hierarchies) |
mdx_ask |
NL question → { status: answer|abstain|clarify, value, mdx, agreement } |
mdx_run |
execute a provided MDX query (read-only) |
mdx_explain |
explain an MDX query in plain language |
Ships a Cube-Analyst persona (persona/) and Claude skills
(skills/) that drive the flow.
Bring your own provider / backend
Implement LLMClient (any model), MdxExecutor (e.g. a Windows ADOMD backend), or Verifier
(e.g. a calibrated trust gate) and inject it — the engine is unchanged.
License
Apache-2.0. No vendor lock-in, no client data.
Recommended Servers
playwright-mcp
A Model Context Protocol server that enables LLMs to interact with web pages through structured accessibility snapshots without requiring vision models or screenshots.
Magic Component Platform (MCP)
An AI-powered tool that generates modern UI components from natural language descriptions, integrating with popular IDEs to streamline UI development workflow.
Audiense Insights MCP Server
Enables interaction with Audiense Insights accounts via the Model Context Protocol, facilitating the extraction and analysis of marketing insights and audience data including demographics, behavior, and influencer engagement.
VeyraX MCP
Single MCP tool to connect all your favorite tools: Gmail, Calendar and 40 more.
graphlit-mcp-server
The Model Context Protocol (MCP) Server enables integration between MCP clients and the Graphlit service. Ingest anything from Slack to Gmail to podcast feeds, in addition to web crawling, into a Graphlit project - and then retrieve relevant contents from the MCP client.
Kagi MCP Server
An MCP server that integrates Kagi search capabilities with Claude AI, enabling Claude to perform real-time web searches when answering questions that require up-to-date information.
E2B
Using MCP to run code via e2b.
Neon Database
MCP server for interacting with Neon Management API and databases
Exa Search
A Model Context Protocol (MCP) server lets AI assistants like Claude use the Exa AI Search API for web searches. This setup allows AI models to get real-time web information in a safe and controlled way.
Qdrant Server
This repository is an example of how to create a MCP server for Qdrant, a vector search engine.