Hive Metadata Exploration

This skill targets AI agent platforms that support terminal command execution (e.g. Claude Code, Qoder).

Safety Red Lines

• Database passwords in the config file must never be printed to the terminal or logs.
• Only read-only exploration operations are performed; Hive metadata is never modified.
• Exploration output files contain no sensitive credentials.
• Never echo the plaintext password from config.ini back to the user.

Tooling Overview

This skill depends on no MCP tools. It connects to the Hive Metastore directly via local Python/Bash scripts.

Prerequisites