The shortest reliable path today is to build the CLI from the Engine repository. This guide uses embedded mode, so it does not start a daemon or require a cluster.

Clone the repository

git clone https://github.com/KrishivAI/krishiv.git
cd krishiv

Build the CLI

cargo build -p krishiv

The first build compiles the Rust workspace and its Arrow/DataFusion dependencies.

Run SQL

cargo run -p krishiv -- sql --query "SELECT 42 AS answer"

The CLI creates an embedded session, plans the query, executes it in process, and prints an Arrow result.

Query a Parquet file

cargo run -p krishiv -- sql \
  --parquet events=./events.parquet \
  --query "SELECT kind, COUNT(*) AS events FROM events GROUP BY kind"

The table=path registration is scoped to that CLI process.

Inspect a plan

cargo run -p krishiv -- explain \
  --query "SELECT 42 AS answer"

Where to go next

Batch SQL guide

Single-node mode

Move the same CLI to a local coordinator and executor.

Python API

Build the native extension from the same source checkout.

Getting started