dolthub/dolt
Dolt is a SQL database with Git-like version control capabilities, enabling users to manage, version, and collaborate on data and schemas as easily as code in Git repositories.
MariaDB/mcp
MCP MariaDB Server is a Model Context Protocol interface to MariaDB databases, maintained in the MariaDB organisation and written in Python. It covers both ordinary SQL work and vector search, so an assistant can explore a schema, run queries and perform semantic retrieval through one server. The standard database tools are list_databases, list_tables, get_table_schema, get_table_schema_with_relations, which adds foreign key relations, execute_sql for SELECT, SHOW and DESCRIBE statements with optional bound parameters, and create_database. Read-only mode is on by default through the MCP_READ_ONLY variable and is enforced inside execute_sql. A second, optional group of tools appears only when an embedding provider is configured: create_vector_store, delete_vector_store, list_vector_stores, insert_docs_vector_store for batch insertion of documents with optional JSON metadata, and search_vector_store for similarity search over the stored embeddings with a configurable number of results. A vector store is an ordinary table with an auto-increment id, a document text column, an indexed embedding column of MariaDB VECTOR type and a JSON metadata column; the distance function defaults to cosine. Supported embedding providers are OpenAI, Gemini and open models from HuggingFace, each selected with EMBEDDING_PROVIDER and the matching API key or model variable. Everything else is configured through environment variables, typically in a .env file: connection host, port, user, password, default database and charset, full SSL and TLS options including CA, client certificate, key and hostname verification, connection pool size, and allowed origins and hosts. The code is organised as server.py for tool definitions, config.py for configuration loading and embeddings.py for the embedding services, with a tests directory. The documentation notes that when the server runs over HTTP or SSE beyond localhost, authentication must be added by the operator, for example using the FastMCP authentication support, since no default method is shipped.
https://github.com/MariaDB/mcp
Also in Databases
t8y2/dbx
DBX is a lightweight, cross-platform database client supporting over 70 databases, featuring a built-in AI assistant, MCP Server, and available as desktop, Docker, or CLI.
Canner/WrenAI
Open-source generative BI engine that exposes governed text-to-SQL, a semantic layer and dashboard generation to AI agents through an MCP server and CLI across more than twenty data sources.
googleapis/mcp-toolbox
MCP Toolbox for Databases is an open-source Model Context Protocol (MCP) server that connects AI agents, IDEs, and applications directly to enterprise databases, offering both pre-built and custom ...
makenotion/notion-mcp-server
The Notion MCP Server is an official MCP server implementation that enables secure, programmatic interaction with Notion content through MCP commands, optimized for AI agents and supporting multiple deployment options.
bytebase/dbhub
DBHub is a universal database gateway implementing the Model Context Protocol (MCP) server interface, enabling MCP clients to connect to and explore multiple database systems including MySQL, PostgreSQL, Oracle, SQL Server, MariaDB, and SQLite.
crystaldba/postgres-mcp
Postgres MCP Pro is an open-source Model Context Protocol server for PostgreSQL that provides advanced database health monitoring, index tuning, query plan analysis, and safe SQL execution to support AI agents and developers throughout the application lifecycle.
supabase-community/supabase-mcp
Supabase MCP Server connects Supabase projects to AI assistants using the Model Context Protocol, enabling LLMs to manage and query Supabase resources.