AI tools · Data & Digital

MCP Toolbox for Databases (genai-toolbox)

Google's open-source MCP server for connecting AI agents to databases, with prebuilt tools, connection pooling, IAM auth and OpenTelemetry. It supports BigQuery, Postgres, MySQL, SQL Server, Oracle, Snowflake, ClickHouse and many more. Aimed at data teams exposing several databases to their agents.

ClickHouse MCP Server

ClickHouse's official MCP server that lets an AI assistant run SQL queries and browse databases and tables on a ClickHouse cluster. It also supports the embedded chDB engine and a read-only mode that guards against destructive operations. Useful for analytics teams working at large scale.

dbt MCP Server

dbt Labs' official MCP server for interacting with dbt Core, dbt Fusion and the dbt Platform. It exposes SQL execution, the Semantic Layer, the Discovery API and CLI commands to AI agents. Aimed at analytics engineering teams modeling their data with dbt.

MotherDuck / DuckDB MCP Server

Official MCP server for DuckDB and MotherDuck that enables SQL analytics over local files, in-memory, S3-hosted and MotherDuck cloud databases. It runs read-only by default. Useful for quickly querying analytical data from an AI assistant.

Databricks MCP Servers (Unity Catalog / Genie)

Databricks Labs MCP servers that expose Unity Catalog data, vector search indexes, functions and Genie spaces to AI agents. They let an assistant query enterprise data. Aimed at data teams working on the Databricks lakehouse.

PostgreSQL MCP server

MCP server that lets Claude query a PostgreSQL database and inspect its schemas directly. The assistant can then answer data questions without you writing SQL by hand. Useful for exploring a relational database in natural language.

Snowflake Managed MCP Server (Cortex)

Snowflake's official managed MCP server: a governed interface that gives AI agents access to Cortex Analyst (structured data via semantic views), Cortex Search (unstructured search) and direct SQL execution, with role-based access control and no separate infrastructure. Aimed at data teams on Snowflake.

Starburst MCP Server (Trino)

MCP server built into Starburst Enterprise/Galaxy: a stateless, authenticated HTTP endpoint for read-only SQL over a Starburst (Trino) cluster, plus data-product search and parameterized query tools. Aimed at teams querying federated data through Trino.

Supabase MCP Server

MCP server that connects an AI assistant to Supabase projects to manage tables, run SQL, handle migrations and query the database. Useful for driving a Supabase Postgres backend from an assistant. Aimed at developers building on Supabase.

Tableau MCP Server

Tableau's official MCP server: a set of tools, resources and prompts that let AI agents query and understand data published in Tableau. Useful for exploring dashboards and data sources from an assistant. Aimed at BI teams on Tableau.

Advanced Data Analysis (docs)

A ChatGPT capability that lets you upload files and spreadsheets, then writes and runs Python to clean, analyze and chart the data. Useful for exploring a dataset and getting results without writing code yourself. Handy for data teams that need quick, ad-hoc analysis.

Data Analyst (GPT)

OpenAI's official Data Analyst GPT: upload a spreadsheet or dataset and it returns analysis, charts and plain-language insights through its code interpreter. Ready to use inside ChatGPT. Useful for getting quick insights with no setup.

Metabase MCP Server (CognitionAI)

Community MCP server covering the Metabase API with 80+ tools: dashboards, cards/questions, databases and tables. It lets an AI assistant run and manage analytics in Metabase. Useful for BI teams that want to drive their dashboards from an assistant.

SQL Chat (GPT)

Custom GPT that helps write, explain and optimize SQL queries, and turns natural-language questions into SQL. Ready to use inside ChatGPT. Useful for analysts and business users who want to build queries without mastering all the syntax.

Qdrant MCP Server

Qdrant's official MCP server that acts as a semantic memory layer: it stores and retrieves vectors through the Qdrant vector search engine for LLM apps. Useful for giving an assistant memory and retrieval. Aimed at teams building RAG applications.

MongoDB MCP Server

MongoDB's official MCP server for MongoDB databases and Atlas clusters. It provides find/aggregate/insert/update/delete plus Atlas cluster and Stream Processing management, in read-only or full-access modes. Aimed at teams working with document data.

Elasticsearch MCP Server

Elastic's official MCP server that connects an AI agent to Elasticsearch indices and lets it query them in natural language. It is now in maintenance mode, superseded by the Elastic Agent Builder MCP endpoint (9.2.0+). Useful for teams running search and data on Elasticsearch.

Chroma MCP Server

Chroma's official MCP server that connects an AI assistant to the Chroma vector database: semantic search, full-text search and metadata filtering across collections. Built for RAG workflows. Aimed at teams building retrieval and memory layers for LLM apps.

Neo4j MCP Server

Neo4j's official MCP server (written in Go) that gives an AI assistant structured access to a Neo4j graph: schema exploration, Cypher execution via natural language and GDS procedure listing. Aimed at teams working with graph data.

Pinecone Developer MCP Server

Pinecone's official MCP server that lets an AI coding assistant search Pinecone docs, manage indexes and query vector data from the development environment. Aimed at developers building semantic search applications.

Jupyter MCP Server

Datalayer's MCP server that lets an AI client control Jupyter notebooks in real time: cell operations, kernel control and live notebook sync. Useful for automating notebook execution and editing. Aimed at data scientists and ML teams.

Vertex AI

Google Cloud's platform for taking Gemini to production: scaling, security, deployment and managed agents. It is used to build, train and run AI models and applications in an enterprise setting. Aimed at data and ML teams on Google Cloud.

LangChain

Widely used open-source framework for building LLM applications and agents across providers (Claude, OpenAI, Gemini). It offers building blocks and abstractions to chain models, tools and data sources. Aimed at developers building AI applications.

LlamaIndex

Open-source data framework focused on RAG that connects your documents and data to any LLM. It provides indexing, retrieval and query tooling to question your sources in natural language. Aimed at developers building retrieval-augmented applications.