Learn more
Discord
🔵
Website
🔵
Issues
A Model Context Protocol (MCP) server implementation for Nile database platform. This server allows LLM applications to interact with Nile platform through a standardized interface.
- Database Management: Create, list, get details, and delete databases
- Credential Management: Create and list database credentials
- Region Management: List available regions for database creation
- SQL Query Support: Execute SQL queries directly on Nile databases
- MCP Protocol Support: Full implementation of the Model Context Protocol
- Type Safety: Written in TypeScript with full type checking
- Error Handling: Comprehensive error handling and user-friendly error messages
- Test Coverage: Comprehensive test suite using Jest
- Environment Management: Automatic loading of environment variables from .env file
- Input Validation: Schema-based input validation using Zod
Install the stable version:
npm install @niledatabase/nile-mcp-server
For the latest alpha/preview version:
npm install @niledatabase/nile-mcp-server@alpha
This will install @niledatabase/nile-mcp-server in your node_modules folder. For example: node_modules/@niledatabase/nile-mcp-server/dist/
# Clone the repository
git clone https://github.com/yourusername/nile-mcp-server.git
cd nile-mcp-server
# Install dependencies
npm install
# Build the project
npm run build
- npx @michaellatman/mcp-get@latest install @niledatabase/nile-mcp-server
There are several ways to start the server:
- Direct Node Execution:
node dist/index.js
- Development Mode (with auto-rebuild):
npm run dev
The server will start and listen for MCP protocol messages. You should see startup logs indicating:
- Environment variables loaded
- Server instance created
- Tools initialized
- Transport connection established
To stop the server, press Ctrl+C
.
When the server starts successfully, you should see logs similar to:
[info] Starting Nile MCP Server...
[info] Loading environment variables...
[info] Environment variables loaded successfully
[info] Creating server instance...
[info] Tools initialized successfully
[info] Setting up stdio transport...
[info] Server started successfully
If you see these logs, the server is ready to accept commands from Claude Desktop.
Create a .env
file in the root directory with your Nile credentials:
NILE_API_KEY=your_api_key_here
NILE_WORKSPACE_SLUG=your_workspace_slug
To create a Nile API key, log in to your Nile account, click Workspaces in the top-left, select your workspace, and navigate to the Security section in the left menu.
- Install Claude Desktop if you haven't already
- Build the project:
npm run build
- Open Claude Desktop
- Go to Settings > MCP Servers
- Click "Add Server"
- Add the following configuration:
{
"mcpServers": {
"nile-database": {
"command": "node",
"args": [
"/path/to/your/nile-mcp-server/dist/index.js"
],
"env": {
"NILE_API_KEY": "your_api_key_here",
"NILE_WORKSPACE_SLUG": "your_workspace_slug"
}
}
}
}
Replace:
/path/to/your/nile-mcp-server
with the absolute path to your project directoryyour_api_key_here
with your Nile API keyyour_workspace_slug
with your Nile workspace slug
- Install Cursor if you haven't already
- Build the project:
npm run build
- Open Cursor
- Go to Settings (⌘,) > Features > MCP Servers
- Click "Add New MCP Server"
- Configure the server:
- Name:
nile-database
(or any name you prefer) - Command:
Replace:
env NILE_API_KEY=your_key NILE_WORKSPACE_SLUG=your_workspace node /absolute/path/to/nile-mcp-server/dist/index.js
your_key
with your Nile API keyyour_workspace
with your Nile workspace slug/absolute/path/to
with the actual path to your project
- Name:
- Click "Save"
- You should see a green indicator showing that the MCP server is connected
- Restart Cursor for the changes to take effect
The server supports two operational modes:
The default mode uses standard input/output for communication, making it compatible with Claude Desktop and Cursor integrations.
Server-Sent Events (SSE) mode enables real-time, event-driven communication over HTTP.
To enable SSE mode:
- Set
MCP_SERVER_MODE=sse
in your.env
file - The server will start an HTTP server (default port 3000)
- Connect to the SSE endpoint:
http://localhost:3000/sse
- Send commands to:
http://localhost:3000/messages
Example SSE usage with curl:
# In terminal 1 - Listen for events
curl -N http://localhost:3000/sse
# In terminal 2 - Send commands
curl -X POST http://localhost:3000/messages \
-H "Content-Type: application/json" \
-d '{
"type": "function",
"name": "list-databases",
"parameters": {}
}'
After setting up the MCP server in Cursor, you can use natural language to interact with Nile databases. Here are some example prompts:
Create a new database named "my_app" in AWS_US_WEST_2 region
List all my databases
Get details for database "my_app"
Delete database "test_db"
Create a users table in my_app database with columns:
- tenant_id (UUID, references tenants)
- id (INTEGER)
- email (VARCHAR, unique per tenant)
- name (VARCHAR)
- created_at (TIMESTAMP)
Create a products table in my_app database with columns:
- tenant_id (UUID, references tenants)
- id (INTEGER)
- name (VARCHAR)
- price (DECIMAL)
- description (TEXT)
- created_at (TIMESTAMP)
Execute this query on my_app database:
SELECT * FROM users WHERE tenant_id = 'your-tenant-id' LIMIT 5
Run this query on my_app:
INSERT INTO users (tenant_id, id, email, name)
VALUES ('tenant-id', 1, 'user@example.com', 'John Doe')
Show me all products in my_app database with price > 100
Show me the schema for the users table in my_app database
Add a new column 'status' to the users table in my_app database
Create an index on the email column of the users table in my_app
The server provides the following tools for interacting with Nile databases:
-
create-database
- Creates a new Nile database
- Parameters:
name
(string): Name of the databaseregion
(string): EitherAWS_US_WEST_2
(Oregon) orAWS_EU_CENTRAL_1
(Frankfurt)
- Returns: Database details including ID, name, region, and status
- Example: "Create a database named 'my-app' in AWS_US_WEST_2"
-
list-databases
- Lists all databases in your workspace
- No parameters required
- Returns: List of databases with their IDs, names, regions, and status
- Example: "List all my databases"
-
get-database
- Gets detailed information about a specific database
- Parameters:
name
(string): Name of the database
- Returns: Detailed database information including API host and DB host
- Example: "Get details for database 'my-app'"
-
delete-database
- Deletes a database
- Parameters:
name
(string): Name of the database to delete
- Returns: Confirmation message
- Example: "Delete database 'my-app'"
-
list-credentials
- Lists all credentials for a database
- Parameters:
databaseName
(string): Name of the database
- Returns: List of credentials with IDs, usernames, and creation dates
- Example: "List credentials for database 'my-app'"
-
create-credential
- Creates new credentials for a database
- Parameters:
databaseName
(string): Name of the database
- Returns: New credential details including username and one-time password
- Example: "Create new credentials for database 'my-app'"
- Note: Save the password when it's displayed, as it won't be shown again
- list-regions
- Lists all available regions for creating databases
- No parameters required
- Returns: List of available AWS regions
- Example: "What regions are available for creating databases?"
- execute-sql
- Executes SQL queries on a Nile database
- Parameters:
databaseName
(string): Name of the database to queryquery
(string): SQL query to executeconnectionString
(string, optional): Pre-existing connection string to use for the query
- Returns: Query results formatted as a markdown table with column headers and row count
- Features:
- Automatic credential management (creates new if not specified)
- Secure SSL connection to database
- Results formatted as markdown tables
- Detailed error messages with hints
- Support for using existing connection strings
- Example: "Execute SELECT * FROM users LIMIT 5 on database 'my-app'"
-
read-resource
- Reads schema information for database resources (tables, views, etc.)
- Parameters:
databaseName
(string): Name of the databaseresourceName
(string): Name of the resource (table/view)
- Returns: Detailed schema information including:
- Column names and types
- Primary keys and indexes
- Foreign key relationships
- Column descriptions and constraints
- Example: "Show me the schema for the users table in my-app"
-
list-resources
- Lists all resources (tables, views) in a database
- Parameters:
databaseName
(string): Name of the database
- Returns: List of all resources with their types
- Example: "List all tables in my-app database"
-
list-tenants
- Lists all tenants in a database
- Parameters:
databaseName
(string): Name of the database
- Returns: List of tenants with their IDs and metadata
- Example: "Show all tenants in my-app database"
-
create-tenant
- Creates a new tenant in a database
- Parameters:
databaseName
(string): Name of the databasetenantName
(string): Name for the new tenant
- Returns: New tenant details including ID
- Example: "Create a tenant named 'acme-corp' in my-app"
-
delete-tenant
- Deletes tenants in the database
- Parameters:
databaseName
(string): Name of the databasetenantName
(string): Name for the tenant
- Returns: Success if the tenant is deleted
- Example: "Delete tenant named 'acme-corp' in my-app"
Here are some example commands you can use in Claude Desktop:
# Database Management
Please create a new database named "my-app" in the AWS_US_WEST_2 region.
Can you list all my databases?
Get the details for database "my-app".
Delete the database named "test-db".
# Connection String Management
Get a connection string for database "my-app".
# Connection string format: postgres://<user>:<password>@<region>.db.thenile.dev:5432/<database>
# Example: postgres://cred-123:password@us-west-2.db.thenile.dev:5432/my-app
# SQL Queries
Execute SELECT * FROM users LIMIT 5 on database "my-app"
Run this query on my-app database: SELECT COUNT(*) FROM orders WHERE status = 'completed'
Using connection string "postgres://user:pass@host:5432/db", execute this query on my-app: SELECT * FROM products WHERE price > 100
All tools return responses in a standardized format:
- Success responses include relevant data and confirmation messages
- Error responses include detailed error messages and HTTP status codes
- SQL query results are formatted as markdown tables
- All responses are formatted for easy reading in Claude Desktop
The server handles various error scenarios:
- Invalid API credentials
- Network connectivity issues
- Invalid database names or regions
- Missing required parameters
- Database operation failures
- SQL syntax errors with helpful hints
- Rate limiting and API restrictions
-
If Claude says it can't access the tools:
- Check that the server path in the configuration is correct
- Ensure the project is built (
npm run build
) - Verify your API key and workspace slug are correct
- Restart Claude Desktop
-
If database creation fails:
- Check your API key permissions
- Ensure the database name is unique in your workspace
- Verify the region is one of the supported options
-
If credential operations fail:
- Verify the database exists and is in the READY state
- Check that your API key has the necessary permissions
nile-mcp-server/
├── src/
│ ├── server.ts # MCP server implementation
│ ├── tools.ts # Tool implementations
│ ├── types.ts # Type definitions
│ ├── logger.ts # Logging utilities
│ ├── index.ts # Entry point
│ └── __tests__/ # Test files
│ └── server.test.ts
├── dist/ # Compiled JavaScript
├── logs/ # Log files directory
├── .env # Environment configuration
├── .gitignore # Git ignore file
├── package.json # Project dependencies
└── tsconfig.json # TypeScript configuration
server.ts
: Main server implementation with tool registration and transport handlingtools.ts
: Implementation of all database operations and SQL query executiontypes.ts
: TypeScript interfaces for database operations and responseslogger.ts
: Structured logging with daily rotation and debug supportindex.ts
: Server startup and environment configurationserver.test.ts
: Comprehensive test suite for all functionality
# Install dependencies
npm install
# Build the project
npm run build
# Start the server in production mode
node dist/index.js
# Start the server using npm script
npm start
# Start in development mode with auto-rebuild
npm run dev
# Run tests
npm test
The following npm scripts are available:
npm run build
: Compiles TypeScript to JavaScriptnpm start
: Starts the server in production modenpm run dev
: Starts the server in development mode with auto-rebuildnpm test
: Runs the test suitenpm run lint
: Runs ESLint for code quality checkingnpm run clean
: Removes build artifacts
The project includes a comprehensive test suite that covers:
- Tool registration and schema validation
- Database management operations
- Connection string generation
- SQL query execution and error handling
- Response formatting and error cases
Run the tests with:
npm test
The server uses structured logging with the following features:
- Daily rotating log files
- Separate debug logs
- JSON formatted logs with timestamps
- Console output for development
- Log categories: info, error, debug, api, sql, startup
MIT License - See LICENSE for details.