Treasure Data MCP Server

Treasure Data MCP Server is an MCP server that connects AI assistants to the Treasure Data customer data platform. Data analysts, engineers, and marketers use it to explore databases, inspect table schemas, and execute SQL queries directly from model interfaces like Claude. By default, the server enforces a read-only environment to safeguard customer data, though write operations can be explicitly enabled through configuration. Beyond raw data warehousing queries, the server includes experimental tools for managing Customer Data Platform segments and activations, allowing assistants to retrieve segment rules and generate underlying audience SQL statements. It also integrates with Treasure Data's Digdag-based workflow engine to inspect workflow projects and monitoring details. The server supports multiple regional endpoints across the United States, Japan, Europe, and Asia-Pacific, while providing automatic row limiting and comprehensive audit logging to keep queries compatible with large language model context windows.

Category: CRM, ERP & E-commerce

Tags: analytics, cdp, customer-data, data-warehousing, treasure-data

Visit Treasure Data MCP Server

How to install and configure Treasure Data MCP Server

1. Ensure Node.js version 18.0.0 or higher is installed on your system. 2. Obtain your Treasure Data API key from your account console. 3. Add the server configuration to your MCP client (such as Claude Desktop) using npx: json { "mcpServers": { "treasuredata": { "command": "npx", "args": ["@treasuredata/mcp-server"], "env": { "TD_API_KEY": "your_api_key", "TD_SITE": "us01", "TD_ENABLE_UPDATES": "false", "TD_DATABASE": "sample_datasets" } } } } Alternatively, if using Claude Code, add it via CLI: bash claude mcp add td -e TD_API_KEY=$TD_API_KEY -- npx @treasuredata/mcp-server 4. Restart your MCP client to start querying Treasure Data.

What you can do with Treasure Data MCP Server

• Exploring databases, listing tables, and inspecting column schemas across datasets without leaving your AI chat client. - Executing read-only SQL queries with automatic row limits and time-range filtering for efficient LLM context consumption. - Auditing customer data platform configurations by inspecting parent segments, sub-segments, and associated activation rules. - Generating underlying SQL definitions for custom customer audience segments to verify targeting criteria. - Monitoring Digdag workflow engine projects and tracking scheduled data processing pipelines directly via conversational commands.

Key facts

• https://github.com/treasure-data/td-mcp-server
• CRM, ERP & E-commerce, Data & Analytics
• analytics, cdp, customer-data, data-warehousing, treasure-data

Part of MCP Servers

What is Treasure Data MCP Server?

Treasure Data MCP Server is a Model Context Protocol server developed by Treasure Data. It allows AI clients such as Claude to interact securely with the Treasure Data customer data platform, enabling schema inspection, SQL queries, CDP audience review, and workflow monitoring.

How do I install Treasure Data MCP Server?

You can run the server directly without manual installation using npx with the package @treasuredata/mcp-server, requiring Node.js 18.0.0 or higher. You configure it in your MCP client by defining the command as npx and passing your TD_API_KEY in the environment variables.

Which MCP clients work with Treasure Data MCP Server?

It works with any client supporting the Model Context Protocol, including Claude Desktop and Claude Code via its built-in MCP integration commands.

Is Treasure Data MCP Server free to use?

During its public preview period, use of the MCP server is free. Treasure Data plans to introduce a usage-based pricing model in the future based on the number of queries issued, with prior notice provided before changes take effect.

Does Treasure Data MCP Server allow data modifications?

By default, the server runs in a secure, read-only mode using the query tool. Write operations like INSERT, UPDATE, and DELETE can only be executed if the TD_ENABLE_UPDATES environment variable is explicitly set to true.