sqlite-utils is a Python command-line tool and library created by Simon Willison (creator of Datasette and Django co-creator) for working with SQLite databases. Available at github.com/simonw/sqlite-utils with extensive documentation at sqlite-utils.datasette.io, it has become a standard tool in the data engineering and analysis toolkit.

The core design principle of sqlite-utils is that SQLite should be as easy to work with as a spreadsheet. You can pipe JSON data from any source directly into a new database and sqlite-utils will automatically create the table with the correct schema. The same works for CSV and TSV files. This makes it trivial to turn any data source — API responses, log files, exported spreadsheets — into a queryable SQLite database in a single command.

One of the most powerful features is in-memory SQL queries against flat files. You can run sqlite-utils memory dogs.csv "select * from t where age > 3" to query a CSV file using SQL without ever creating a database file. This extends to joins across multiple files, aggregations, and complex queries. It supports full-text search configuration, allowing you to index text columns and run relevance-ranked search queries.

For schema management, sqlite-utils provides transformation commands that work around SQLite’s limited ALTER TABLE support. You can change column types, rename columns, reorder columns, and extract normalized tables from denormalized data — operations that would normally require creating a new table, copying data, and dropping the old one.

A skill built on sqlite-utils gives an AI agent a fast, flexible way to create, query, and transform structured data. The agent could ingest API responses into local databases, run analytical queries, normalize messy data, or build full-text search indexes. Install via pip install sqlite-utils or brew install sqlite-utils. The tool supports plugins for custom SQL functions and is Apache 2.0 licensed.