Skip to content

Getting Started

Installation

No external servers needed — embedded ClickHouse and SQLite included:

pip install aaiclick
python -m aaiclick setup

For a remote ClickHouse server and PostgreSQL:

pip install "aaiclick[distributed]"
python -m aaiclick setup
export AAICLICK_CH_URL="clickhouse://user:pass@host:8123/db"
export AAICLICK_SQL_URL="postgresql+asyncpg://user:pass@host:5432/db"

Add lineage tracing and debug agents:

pip install "aaiclick[ai]"
# or everything at once:
pip install "aaiclick[all]"

Quick Example

import asyncio
from aaiclick import create_object_from_value
from aaiclick.data.data_context import data_context

async def main():
    async with data_context():
        prices = await create_object_from_value([10.0, 20.0, 30.0])

        total = prices + prices * 0.1               # LazyOperator — no DB call yet
        print(await total.data())                   # [11.0, 22.0, 33.0]
        print(await total.mean().data())            # 22.0

asyncio.run(main())

Always await operation results

Forgetting await causes confusing errors downstream, not at the forgotten line.

Environment Variables

Variable Default Description
AAICLICK_CH_URL chdb:///~/.aaiclick/chdb_data ClickHouse connection — chdb:// for embedded, clickhouse:// for remote
AAICLICK_SQL_URL sqlite+aiosqlite:///~/.aaiclick/local.db Orchestration DB — SQLite (local) or PostgreSQL (distributed)
AAICLICK_LOG_DIR ~/.aaiclick/logs / /var/log/aaiclick Log directory override (macOS default / Linux default)

Next Steps

  • Object API — operators, aggregations, views, group by
  • DataContext — lifecycle management, persistent objects
  • Orchestration — @task and @job decorators, workers
  • Examples — runnable scripts for every feature