This embeds ClickHouse's SQL engine directly in your Python process so you can query parquet, CSV, JSON files, S3 buckets, and remote databases without spinning up a server. The appeal is running analytical queries with ClickHouse's 1000+ functions (window functions, geo functions, JSON operators) on whatever data source you point it at, including cross-source joins between a Postgres table and a local parquet file in one query. Use the simple query() function for one-offs or Session for multi-step pipelines that persist tables. It's Apache 2.0 licensed and requires Python 3.9+ on macOS or Linux. Skip this if you want pandas-style DataFrame operations; there's a separate chdb-datastore skill for that.
npx skills add https://github.com/clickhouse/agent-skills --skill chdb-sql