Verodat MCP Server
Overview
A Model Context Protocol (MCP) server implementation for Verodat, enabling seamless integration of Verodat's data management capabilities with AI systems like Claude Desktop.
Verodat MCP Server
This repository contains a Model Context Protocol (MCP) server implementation for Verodat, allowing AI models to interact with Verodat's data management capabilities through well-defined tools.
Overview
The Verodat MCP Server provides a standardized way for AI models to access and manipulate data in Verodat. It implements the Model Context Protocol specification, providing tools for data consumption, design, and management.
Tool Categories
The server is organized into three main tool categories, each offering a progressive set of capabilities:
1. Consume (8 tools)
The base category focused on data retrieval operations:
get-accounts: Retrieve available accountsget-workspaces: List workspaces within an accountget-datasets: List datasets in a workspaceget-dataset-output: Retrieve actual data from a datasetget-dataset-targetfields: Retrieve field definitions for a datasetget-queries: Retrieve existing AI queriesget-ai-context: Get workspace context and data structureexecute-ai-query: Execute AI-powered queries on datasets
2. Design (9 tools)
Includes all tools from Consume, plus:
create-dataset: Create a new dataset with defined schema
3. Manage (10 tools)
Includes all tools from Design, plus:
upload-dataset-rows: Upload data rows to existing datasets
Prerequisites
- Node.js (v18 or higher)
- Git
- Claude Desktop (for Claude integration)
- Verodat account and AI API key
Installation
Quick Start
Installing via Smithery
To install Verodat MCP Server for Claude Desktop automatically via Smithery:
npx -y @smithery/cli install @Verodat/verodat-mcp-server --client claude
Manual Installation
- Clone the repository:
git clone https://github.com/Verodat/verodat-mcp-server.git
cd verodat-mcp-server
- Install dependencies and build:
npm install
npm run build
-
Configure Claude Desktop: Create or modify the config file:
- MacOS:
~/Library/Application Support/Claude/claude_desktop_config.json - Windows:
%APPDATA%/Claude/claude_desktop_config.json
Add the configuration which is mensioned below in configuration:
- MacOS:
Getting Started with Verodat
- Sign up for a Verodat account at verodat.com
- Generate an AI API key from your Verodat dashboard
- Add the API key to your Claude Desktop configuration
Configuration
The server requires configuration for authentication and API endpoints. Create a configuration file for your AI model to use:
{
"mcpServers": {
"verodat-consume": {
"command": "node",
"args": [
"path/to/verodat-mcp-server/build/src/consume.js"
],
"env": {
"VERODAT_AI_API_KEY": "your-api-key",
"VERODAT_API_BASE_URL": "https://verodat.io/api/v3"
}
}
}
}
Configuration Options
You can configure any of the three tool categories by specifying the appropriate JS file one at a time in claude:
- Consume only: Use
consume.js(8 tools for data retrieval) - Design capabilities: Use
design.js(9 tools, includes dataset creation) - Full management: Use
manage.js(10 tools, includes data upload)
Example for configuring all three categories simultaneously:
{
"mcpServers": {
"verodat-consume": {
"command": "node",
"args": [
"path/to/verodat-mcp-server/build/src/consume.js"
],
"env": {
"VERODAT_AI_API_KEY": "your-api-key",
"VERODAT_API_BASE_URL": "https://verodat.io/api/v3"
}
},
"verodat-design": {
"command": "node",
"args": [
"path/to/verodat-mcp-server/build/src/design.js"
],
"env": {
"VERODAT_AI_API_KEY": "your-api-key",
"VERODAT_API_BASE_URL": "https://verodat.io/api/v3"
}
},
"verodat-manage": {
"command": "node",
"args": [
"path/to/verodat-mcp-server/build/src/manage.js"
],
"env": {
"VERODAT_AI_API_KEY": "your-api-key",
"VERODAT_API_BASE_URL": "https://verodat.io/api/v3"
}
}
}
}
Environment Variables
VERODAT_AI_API_KEY: Your Verodat API key for authenticationVERODAT_API_BASE_URL: The base URL for the Verodat API (defaults to "https://verodat.io/api/v3" if not specified)
Tool Usage Guide
Available Commands
The server provides the following MCP commands:
// Account & Workspace Management
get-accounts // List accessible accounts
get-workspaces // List workspaces in an account
get-queries // Retrieve existing AI queries
// Dataset Operations
create-dataset // Create a new dataset
get-datasets // List datasets in a workspace
get-dataset-output // Retrieve dataset records
get-dataset-targetfields // Retrieve dataset targetfields
upload-dataset-rows // Add new data rows to an existing dataset
// AI Operations
get-ai-context // Get workspace AI context
execute-ai-query // Run AI queries on datasets
Selecting the Right Tool Category
- For read-only operations: Use the
consume.jsserver configuration - For creating datasets: Use the
design.jsserver configuration - For uploading data: Use the
manage.jsserver configuration
Security Considerations
- Authentication is required via API key
- Request validation ensures properly formatted data
Development
The codebase is written in TypeScript and organized into:
- Tool handlers: Implementation of each tool's functionality
- Transport layer: Handles communication with the AI model
- Validation: Ensures proper data formats using Zod schemas
Debugging
The MCP server communicates over stdio, which can make debugging challenging. We provide an MCP Inspector tool to help:
npm run inspector
This will provide a URL to access debugging tools in your browser.
Contributing
We welcome contributions! Please feel free to submit a Pull Request.
License
LICENSE file for details
Support
- Documentation: Verodat Docs
- Issues: GitHub Issues
- Community: Verodat Community