Nile MCP Server

What niledatabase/nile-mcp-server MCP server does

The niledatabase/nile-mcp-server MCP server connects MCP-compatible applications to Nile’s database platform. It provides tools for managing databases in a Nile workspace, including creating a database, listing existing databases, retrieving database details, and deleting a database.

Database creation accepts a database name and a supported region. The README identifies AWS_US_WEST_2 and AWS_EU_CENTRAL_1 as available region values. Database details can include identifiers, status, API host, and database host.

The server also handles database credentials. An agent can list credentials for a named database or create new credentials. Newly created credentials include a username and one-time password; the password must be saved when displayed because it will not be shown again.

SQL execution is another supported use case. Agents can send SQL statements to a selected Nile database, making the server suitable for querying data as well as performing schema or data changes when the connected database and credentials permit them.

How it works

The niledatabase/nile-mcp-server MCP server is implemented in TypeScript and communicates using the Model Context Protocol. Its default transport is stdio, which is the mode documented for Claude Desktop and Cursor. The project uses schema-based input validation and reports errors through user-facing messages.

An optional SSE mode starts an HTTP service, using port 3000 by default. Clients can listen on /sse and submit commands to /messages. This mode is useful when the MCP connection needs an HTTP event stream rather than a local process connected through standard input and output.

Setup and configuration

Install the npm package or clone the repository, install dependencies, and build the project. The documented local process starts with node dist/index.js; development mode is available through the repository’s npm run dev script.

Set these values in a .env file or the client configuration:

• NILE_API_KEY: an API key created from the Nile workspace security settings.
• NILE_WORKSPACE_SLUG: the Nile workspace to access.

Claude Desktop and Cursor configurations invoke the built dist/index.js file and pass the credentials through environment variables. Absolute paths are required in those client configurations. To use SSE instead of stdio, set MCP_SERVER_MODE=sse; the server then exposes the documented local HTTP endpoints.

Tools and capabilities

The niledatabase/nile-mcp-server MCP server supports these operation groups:

• Create, list, inspect, and delete Nile databases.
• List available database regions.
• Create and list credentials for a database.
• Execute SQL against a selected database.
• Use MCP stdio transport or opt into SSE transport.

The README examples also show natural-language workflows for creating tables, inspecting table schemas, adding columns, creating indexes, and querying or modifying rows through SQL.

Limitations and notes

The documented region choices are limited to the two values listed above. Database deletion is exposed as an agent operation, so client users should treat destructive requests carefully. SQL statements can change data or schema depending on the query supplied.

The niledatabase/nile-mcp-server MCP server requires Nile credentials and a workspace slug. The provided material documents Claude Desktop and Cursor setup, but it does not establish compatibility with other clients. It also does not state a project license or a hosted Nile MCP endpoint.