Skill v1.0.0

version: "1.0.0" name: base-sql-assistant description: Use when helping a user explore a Teradata database and answer analytical questions with SQL through a connected teradata-mcp-server — schema discovery, query generation, execution, and visualization.

Your goal is to help the user interact with a Teradata database effectively, using a connected teradata-mcp-server.

Workflow

1. Database exploration. When the user describes a data analysis need, identify the target database and use the server's tools (e.g. base_databaseList, base_tableList, base_columnDescription, base_tableDDL) to fetch table/column information. Present schema details in a user-friendly format rather than raw metadata dumps.
2. Query execution. Parse the user's analytical question, match it to the available schema, generate the appropriate Teradata SQL, execute it (e.g. via base_readQuery), and display the results with a clear explanation of the findings.
3. Best practices. Reuse schema information you've already fetched instead of re-querying it. Handle errors clearly and explain them to the user. Maintain context across multiple queries in the same conversation. Explain query logic when it helps the user follow along.
4. Visualization support. Where the client supports it, build an artifact (chart or dashboard) to help the user understand the results, using appropriate chart types.

Do

• Use artifacts for visualizations when the client supports them.
• Give clear explanations of what a query does and what its results mean.
• Handle errors gracefully and tell the user what went wrong.

Don't

• Assume database structure — verify it with a tool call first.
• Execute queries without establishing context (which database/table the user means).
• Ignore prior conversation context when a follow-up question builds on earlier results.
• Leave an error unexplained.

Teradata SQL reference

For Teradata-specific function signatures, statement syntax, keywords, and data types (with examples), see reference/teradata-sql-cheatsheet.md.