</>MCP Agents Market
Skill

chdb

by chdb-io2.9kPythonUpdated 2026-09-04

chDB is an in-process OLAP SQL Engine πŸš€ powered by ClickHouse

CursorClaude CodeCodexWindsurf

chDB is an in-process OLAP SQL engine powered by ClickHouse that can be integrated into AI coding agents through dedicated agent skills. It provides two distinct skillsβ€”chdb-datastore (a pandas-compatible API) and chdb-sql (direct SQL querying)β€”that enable AI assistants like Cursor, Claude Code, and Codex to generate correct chdb code out of the box. The skills allow agents to perform high-performance analytics on Parquet, CSV, JSON, Arrow files, and databases without installing a separate server. chDB also supports WebAssembly execution, UDFs, streaming queries, and direct integration with frameworks like LlamaIndex, Pydantic AI, and MCP.

Key Features

Two AI agent skills: chdb-datastore (pandas drop-in replacement) and chdb-sql (in-process SQL queries)
In-process ClickHouse OLAP engine requiring no server installation
Pandas-compatible DataStore API with lazy evaluation and automatic SQL generation
Supports 60+ file formats including Parquet, CSV, JSON, Arrow, and ORC
WebAssembly build (chDB-Wasm) runs SQL directly in browsers with zero install
Query data from S3, MySQL, PostgreSQL, SQLite, and other remote sources
Python DB-API 2.0 support with stateful sessions and streaming queries
Integrations with LlamaIndex, Pydantic AI, smolagents, MCP, and Vercel AI SDK

Use Cases

  • 01Teach AI coding assistants to generate high-performance data analysis code using chdb
  • 02Enable agents to query large datasets (Parquet, CSV, JSON) with ClickHouse performance without server setup
  • 03Replace pandas in existing AI-generated code to achieve SQL-level speed on large data
  • 04Build AI-powered data pipelines that query S3, databases, and local files using natural language
  • 05Create browser-based SQL analytics tools using the WebAssembly build
  • 06Integrate ClickHouse analytics capabilities into agent frameworks like LlamaIndex and Pydantic AI

Related Skills

View more

chdb β€” FAQ

What are the chdb AI agent skills?+

chDB provides two agent skills: chdb-datastore, which teaches agents to use the pandas-compatible DataStore API, and chdb-sql, which teaches in-process SQL querying. These skills enable AI coding assistants to generate correct chdb code automatically.

How do I install the chdb agent skills?+

Run the installation script with 'curl -sL https://raw.githubusercontent.com/chdb-io/chdb/main/install_skill.sh | bash' to install both skills for your AI assistant. For project-level installation (shared with your team), add the '--project' flag.

Which AI assistants work with chdb skills?+

chDB skills work with Cursor, Claude Code, Codex, Windsurf, and other AI coding agents that support skill/capability systems. The package also integrates with LlamaIndex, Pydantic AI, smolagents, and MCP-compatible agents.

Do I need API keys or a ClickHouse server?+

No API keys or ClickHouse server installation are required. chDB runs entirely in-process. The optional AI-assisted SQL generation feature requires an OpenAI or Anthropic API key, configured via environment variables.

Is chdb free to use?+

Yes, chDB is open source under the Apache 2.0 license. The Python package can be installed via pip for free, and the WebAssembly version runs in browsers with no cost.

What file formats and data sources does chdb support?+

chDB supports 60+ formats including Parquet, CSV, JSON, Arrow, and ORC. It can query local files, S3 buckets, MySQL, PostgreSQL, SQLite, MongoDB, ClickHouse, HDFS, Azure, and GCS without data movement.

How do I install chdb?+

Open the source repository on GitHub and follow its README. chdb is a skill β€” MCP Agents Market links you directly to the official repo.

Is chdb free?+

chdb is an open-source project hosted on GitHub. Check the repository for its license and any usage requirements.

Related searches