DuckDB CLI skills

**Quick judgment**: This skill is a strong choice whenever you want an AI agent to run serious SQL-based analytics or file conversions on local tabular data using DuckDB. It exposes much of the DuckDB CLI’s power, including multi-format I/O, globbing, and useful output modes, making it well-suited for data-heavy workflows. **Key strengths** - High analytical capability via DuckDB’s SQL engine over CSV/Parquet/JSON and DuckDB DB files. - Robust data conversion support (CSV↔Parquet, JSON→Parquet, filtered exports) directly from SQL queries. - Flexible output formats (CSV, JSON, markdown, HTML, LaTeX, INSERT statements) for both human inspection and downstream tooling. - Good fit for scripting and automation with -c/-f, stdin/stdout, and dot commands for configuration. **Risks and limitations** - **File access/global data exposure**: DuckDB CLI can read and write arbitrary accessible files unless constrained; in unconstrained environments this can expose sensitive data or overwrite existing files. - **Resource usage**: Large datasets or complex joins may be memory- and CPU-intensive; queries without LIMIT or proper filtering could be slow or expensive. - **SQL dependence**: Effective use requires reasonable SQL literacy; for non-technical users the agent may need careful prompting or templates. - **Environment dependency**: Requires DuckDB CLI to be installed and properly available; behavior may differ across OSes or container setups. **Recommended scenarios** - Letting an AI agent perform **local data analysis and summarization** on CSV/Parquet/JSON logs, exports, or analytics datasets. - Implementing **on-the-fly data conversion** pipelines (e.g., CSV imports to Parquet for faster repeated querying). - Running **reproducible analytical scripts** or reports where the agent orchestrates DuckDB queries and exports results in a specific format. - Using a lightweight, embedded OLAP engine instead of standing up a separate database service for analytical workloads.