Overview
Verodat MCP Server is an implementation of the Model Context Protocol (MCP) for Verodat, enabling AI models to access and interact with Verodat’s data management capabilities through a standardized set of tools. It provides a structured interface that organizes operations into three tool categories—Consume, Design, and Manage—each unlocking progressively more advanced data workflows. Consume offers eight read-oriented tools for retrieving accounts, workspaces, datasets, dataset outputs, dataset target fields, AI queries, workspace AI context, and executing AI queries. Design adds the ability to create datasets with defined schemas, and Manage adds uploading dataset rows. The server is designed to be integrated with Claude Desktop, with configuration via a JSON file that maps mcpServers to scripts (consume.js, design.js, manage.js) and environment variables VERODAT_AI_API_KEY and VERODAT_API_BASE_URL. Prerequisites include Node.js v18+, Git, a Verodat account and API key, and Claude Desktop. The codebase is TypeScript-based, with distinct layers for tool handlers, transport, and validation using Zod, and provides an MCP Inspector for debugging via npm run inspector that exposes a browser-accessible URL.
Features
get-accounts
Fetch the list of accounts the user can access in Verodat.
get-workspaces
List workspaces within an account to scope data operations.
get-datasets
List datasets in a workspace to identify available data assets.
get-dataset-output
Retrieve dataset records/outputs for use in AI workflows.
get-dataset-targetfields
Retrieve dataset target field definitions to guide queries.
get-queries
Retrieve existing AI queries for reuse and auditing.
get-ai-context
Get workspace AI context and data structure information.
execute-ai-query
Run AI-powered queries on datasets to generate insights.
Who Is This For?
- AI models:Access Verodat data and run MCP tools within Claude Desktop integration.
