chDB Datastore
Data & Analyticsby clickhouse
3/3 audits passApache-2.0
Use when the user has tabular data (pandas DataFrame, parquet, csv, Arrow, json) and wants to filter, group, aggregate, join, or speed up slow pandas.
chDB Datastore, published by ClickHouse under Apache-2.0, teaches the agent to use chDB DataStore: a lazy, ClickHouse-backed replacement for pandas. The central idea is changing one import (import chdb.datastore as pd) and keeping existing pandas code, with operations compiled to SQL and executed only when results are needed. A decision tree routes the agent between loading a file or database, joining data across sources and speeding up slow pandas code, and sends raw SQL work to a separate chdb-sql skill.
It shows one connection pattern for local files (parquet, csv, json, arrow and more), MySQL, PostgreSQL, S3, MongoDB, ClickHouse, Iceberg and Delta Lake, then the familiar pandas operations: filtering, groupby, assign, string and datetime accessors, and joins, including joining a MySQL table to a parquet file. It covers writing results back out, using to_sql() to inspect generated SQL, and a troubleshooting table. It requires Python 3.9 or later on macOS or Linux, and ships reference files, examples and a verify_install.py script.
This is mainly a skill for coding agents such as Claude Code working in Python; if imported into BusinessMCP only its instructions come across, and the bundled script is not run.
What you can do with it
- Speed up slow pandas code by changing a single import
- Join a MySQL table with a local parquet file in one DataFrame
- Read S3, Iceberg or Delta Lake data as a DataFrame
- Inspect the SQL a DataFrame pipeline compiles to when debugging
Run it on your business data
Imported into BusinessMCP, chDB Datastore becomes a playbook your AI business analyst applies to your web analytics, funnels and connected databases.
Use chDB Datastore in BusinessMCPInstall it in a coding agent
One command adds chDB Datastore to your project.
npx skills add https://github.com/clickhouse/agent-skills --skill chdb-datastoreHow we vetted it
- Source
- clickhouse/agent-skills at 2f6ec4b
- Licence
- Apache-2.0
- Security audits (skills.sh)
- Gen Agent Trust Hub: Pass · Socket: Pass · Snyk: Pass
- Bundled scripts
- Yes — read them before installing
Checked 2026-09-25 against its skills.sh listing. How we vet skills
Related skills
All skillsJupyter Notebook
Use when the user asks to create, scaffold, or edit Jupyter notebooks (`.ipynb`) for experiments, explorations, or tutorials;
SQL Queries
Write correct, performant SQL across all major data warehouse dialects (Snowflake, BigQuery, Databricks, PostgreSQL, etc.).
Explore Data
Profile and explore a dataset to understand its shape, quality, and patterns. Use when encountering a new table or file, checking null rates and column distributions, spotting data quality issues like duplicates or…
Frequently asked questions
What is chDB DataStore?
A pandas-compatible API backed by the ClickHouse engine. Operations are lazy, compile to SQL and run only when a result is needed, such as on print or len.
What are the requirements?
Python 3.9 or later on macOS or Linux, with chdb installed via pip. The skill ships a verify_install.py script to check the environment.
How do I install chDB Datastore?
Run `npx skills add clickhouse/agent-skills --skill chdb-datastore`, or import it from the BusinessMCP dashboard as a playbook (instructions only; the bundled script is not run).
Keep exploring
Run chDB Datastore against your whole business
Free plan, no credit card.
Use chDB Datastore in BusinessMCP