Guides8 min read

Best MCP Servers for Data Science & Analytics in 2026

Unlock AI-powered data science workflows with MCP servers for Jupyter notebooks, BigQuery, Databricks, dbt, and more. Stop context-switching and let your AI work directly with your data infrastructure.

By MyMCPTools Team·

Data science workflows involve constant context switching: you're in Jupyter, then BigQuery, then dbt, then Slack explaining what you found. MCP servers collapse this stack. Your AI assistant can query your data warehouse, run notebook cells, check pipeline health, and explain results — all in one continuous conversation.

What Changes When Data Scientists Use MCP

The traditional AI coding assistant model requires you to paste code snippets and results into chat. MCP flips this: your AI becomes an active participant in your data environment, capable of reading live data, executing queries, and iterating on analysis in real time.

1. Jupyter MCP Server — AI-Assisted Notebooks

The Jupyter MCP server gives your AI direct access to running Jupyter notebooks — reading cells, executing code, and inspecting outputs without copy-pasting. This is the closest thing to a genuine AI pair programmer for data science.

Key capabilities:

  • Read and execute notebook cells directly
  • Access variable state and dataframe previews
  • Inspect outputs, errors, and visualizations
  • Create new cells and modify existing ones

Best for: Data scientists doing exploratory analysis who want AI collaboration without copy-pasting code blocks back and forth. Works with JupyterLab and classic Jupyter.

2. BigQuery MCP Server — SQL at Google Scale

BigQuery processes petabytes. The BigQuery MCP server gives your AI the ability to write, execute, and explain SQL queries directly against your BigQuery datasets — no manual query copying required.

Key capabilities:

  • Execute SQL queries and return results
  • Describe table schemas and dataset structure
  • Estimate query costs before execution
  • Access query history and saved queries
  • Create and manage tables programmatically

Best for: Data analysts and engineers working in GCP environments who want AI-assisted SQL generation and optimization at scale.

3. Databricks MCP Server — Unified Analytics Platform

Databricks is the dominant platform for large-scale data engineering and ML workloads. Its MCP server connects your AI to Databricks clusters, notebooks, Delta tables, and Unity Catalog — making enterprise data accessible through natural language.

Key capabilities:

  • Query Delta tables and Unity Catalog assets
  • Run Spark SQL and Python code in Databricks notebooks
  • Monitor cluster health and job runs
  • Access ML experiment tracking (MLflow integration)

Best for: Enterprise data teams running large-scale ETL pipelines, feature engineering, and ML training on Databricks.

4. dbt MCP Server — Data Transformation Workflows

dbt has become the standard for analytics engineering. The dbt MCP server lets your AI understand your transformation models, run them, check test results, and help debug lineage issues — turning dbt from a command-line tool into an AI-collaborative environment.

Key capabilities:

  • Parse and explain dbt model definitions
  • Run dbt commands (run, test, compile, docs)
  • Inspect model lineage and dependencies
  • Access test results and failure details

Best for: Analytics engineers managing dbt projects who want AI assistance for model development, debugging, and documentation.

5. Apache Spark MCP Server — Distributed Processing

For truly large-scale data processing, the Apache Spark MCP server bridges your AI with Spark clusters. Submit jobs, monitor execution plans, and debug performance issues with AI assistance.

Key capabilities:

  • Submit and monitor Spark jobs
  • Inspect execution plans and query optimizations
  • Access Spark UI metrics programmatically
  • Read partitioned dataset schemas and metadata

Best for: Data engineers running large-scale batch processing on Spark clusters (AWS EMR, Google Dataproc, Azure HDInsight).

6. Excel MCP Server — Spreadsheet Intelligence

Not every data science team works with petabytes. The Excel MCP server brings AI assistance to the world's most widely used data tool — reading sheets, running formulas, and helping analysts who live in spreadsheets.

Key capabilities:

  • Read and write Excel files (.xlsx, .xls, .csv)
  • Execute formulas and return computed values
  • Analyze data ranges and suggest pivot configurations
  • Handle multi-sheet workbooks

Best for: Business analysts, financial modelers, and data professionals who primarily work in Excel and want AI assistance without migrating to code-first tools.

7. Google Analytics MCP Server — Web Data Access

For data teams responsible for web analytics, the Google Analytics MCP server enables natural language querying of GA4 data — no more navigating GA's complex exploration interface to pull basic metrics.

Key capabilities:

  • Query GA4 dimensions and metrics through natural language
  • Pull traffic, conversion, and engagement reports
  • Compare date ranges and segments
  • Export data for downstream analysis

Best for: Digital analytics teams, growth engineers, and marketing data analysts standardized on GA4.

The Data Science AI Stack

Build your stack based on your environment:

  • Notebook-first teams: Jupyter MCP + Filesystem MCP + BigQuery/Databricks MCP
  • Analytics engineering teams: dbt MCP + BigQuery/Databricks MCP + GitHub MCP
  • Enterprise Spark shops: Databricks MCP + Apache Spark MCP + Git MCP
  • Spreadsheet-centric teams: Excel MCP + Google Analytics MCP + Filesystem MCP

The right combination turns your AI into a data teammate that can actually run queries, debug pipelines, and explain results — not just suggest code you have to execute yourself.

Related guides:

Recommended Tools

Better Stack

Free Plan

Get alerted when your APIs, browser tests, payment pipelines, or MCP server dependencies go down. Used by 100K+ developers.

Start monitoring free →

1Password

14-day Free Trial

Store and inject API keys, payment credentials, tokens, and file access secrets into your MCP server configs. Trusted by 150K+ developers.

Try 1Password free →

🔧 MCP Servers Mentioned in This Article

💻

JupyterLab

Control JupyterLab notebooks from AI assistants. Execute cells, inspect variables, visualize outputs, and manage kernels programmatically.

Local
🗄️

BigQuery MCP Server

Google's official BigQuery MCP integration ships as part of the MCP Toolbox for Databases (googleapis/mcp-toolbox, 15,800+ stars, formerly published under the genai-toolbox repo name before Google renamed it), a single Go-based server binary that speaks the Model Context Protocol for over a dozen Google Cloud and third-party databases. Rather than a BigQuery-only package, you run the shared toolbox binary with a `--prebuilt=bigquery` flag to instantly load BigQuery-specific tools — schema/table discovery (`list_dataset_ids`, `list_table_ids`, `get_table_info`), running arbitrary SQL via `execute_sql`, and dry-run query validation for cost estimation before executing — over stdio or as an HTTP/SSE server. The quickest install is `npx -y @toolbox-sdk/server --prebuilt=bigquery --stdio` in your MCP client config; it also ships as a downloadable binary and Docker image for teams that prefer not to run via npx. Authentication uses standard Google Cloud credential chains (Application Default Credentials, service account keys, or Workload Identity) rather than embedding a project-specific key. Toolbox also underlies Google's official SDKs for Python, JS/TS, Go, and Java, so the same server config can back both ad hoc AI-assistant queries ("show me the schema for the events table and the row count for last week") and production agent tools built with LangChain, LlamaIndex, or ADK. For teams that want a fully managed remote option instead of self-hosting, Google Cloud also offers managed MCP servers for its databases including BigQuery.

Local
🗄️

Databricks

Connect to data, AI tools & agents, and the rest of the Databricks platform using turnkey managed MCP servers.

Local
🗄️

dbt (data build tool)

Transform data in your warehouse with dbt. Run models, test assertions, generate docs, inspect lineage, and manage dbt Cloud jobs via MCP.

Local
🗄️

Apache Spark

Unified analytics engine MCP for Apache Spark. Submit jobs, query DataFrames via Spark SQL, inspect execution plans, and analyze large-scale distributed data.

Local
📁

Excel MCP Server

Excel MCP Server (by haris-musa, nearly 4,000 GitHub stars) lets AI agents create, read, and edit Excel workbooks without Microsoft Excel installed anywhere in the pipeline. It's a Python-based server exposing tools across the full spreadsheet lifecycle: creating and modifying workbooks and worksheets, writing formulas, building and styling Excel Tables, generating charts (line, bar, pie, scatter, and more), constructing dynamic pivot tables for analysis, and applying rich formatting — fonts, colors, borders, alignment, and conditional formatting rules. Built-in data validation keeps ranges, formulas, and cell contents consistent as an agent edits a file. The server supports three transports: stdio for local single-user setups (the default for Claude Desktop and Claude Code), plus SSE and streamable HTTP for remote deployments — when running remotely, set the EXCEL_FILES_PATH environment variable so the server knows where to read and write files, and FASTMCP_PORT to control the listening port. This makes it equally useful for a solo analyst automating a weekly report locally and for a team running a shared Excel-manipulation service that multiple agents call into. Because it operates on the raw XLSX format directly, there's no licensing dependency on Excel itself, and workflows like "pull this CSV into a formatted table with a pivot summary and a bar chart" become a single natural-language request instead of a manual multi-step process.

Local
📊

Google Analytics

Query Google Analytics 4 data via MCP. Analyze traffic, user behavior, conversions, and audience segments using GA4's reporting API.

Local

📚 More from the Blog