teradata

The Teradata MCP Server provides a collection of tools and prompts organized into modules to help agents interact with Teradata databases. It exposes a suite of toolkits for data access, querying, data quality checks, and administration tasks, enabling AI agents and users to search vector stores, query the Teradata platform, access feature stores, run data quality checks, and manage DBA responsibilities. The server is designed to be invoked via an MCP client (for example Claude Desktop or other MCP clients) and can be configured with a DATABASE_URI to connect to your Teradata instance. After starting, you can leverage the pre-defined tool groups and prompts to build workflows, run queries, and orchestrate data tasks across Teradata components.

Prerequisites:

• Python 3.8+ (recommended)
• Access to Teradata environment or sandbox for testing
• Optional: uv (for running the MCP server via uvx as shown in the Quick Start)

Installation steps:

1. Install the MCP server package from PyPI (choose one of the two approaches):

Option A: Install with pipx (recommended for isolated environments)

pipx install teradata-mcp-server

Option B: Install with pip (global or virtualenv)

python3 -m pip install --user teradata-mcp-server

2. Verify installation

python -m teradata_mcp_server --version

(If installed via pipx, the command may be teradata-mcp-server --version.)

3. Prepare configuration for uvx usage as described in the Quick Start. You will typically set an environment URI that points to your Teradata instance, for example:

DATABASE_URI=teradata://<USERNAME>:<PASSWORD>@<HOST_URL>:1025/<USERNAME>

4. Run the MCP server using uvx (per the Quick Start instructions):

uvx teradata-mcp-server

Notes:

• If you prefer not to use uvx, you can also run a direct Python invocation depending on your setup, but the recommended path for Claude Desktop users is via uvx.
• Ensure network access from the host running the MCP server to your Teradata systems.

Tips and common issues:

• Ensure the DATABASE_URI environment variable is correctly formatted and reachable from the MCP server host.
• If you update the teradata-mcp-server package, re-run the installation steps to refresh capabilities.
• The MCP server exposes modules for Search, Query, Table, Data Quality, DBA, Data Scientist, and BAR tools. Explore the docs for each module to understand supported prompts and prompts usage.
• When using Claude Desktop or other MCP clients, configure the client to point to the running MCP server using the appropriate mcpServers configuration (as shown in the Quick Start example).
• If you encounter connectivity errors, verify firewall rules, DNS resolution, and that the Teradata URI uses the correct port and credentials.