exasol-quickstart 0.3.4

One command to try Exasol with AI add-ons (MCP Server + JSON Tables) — auto-selects the right setup for your OS (Windows / macOS / Linux).

exasol-quickstart

Run Exasol with AI add-ons in one command — the database, an MCP server, and JSON Tables, set up the right way for your operating system.

PyPI · GitHub

---

Quickstart

To try it — runs once, nothing left installed:

pipx run exasol-quickstart      # with pipx
uvx exasol-quickstart           # with uv

To keep it — installs the command so you can run it again later:

pipx install exasol-quickstart && exasol-quickstart      # with pipx
uv tool install exasol-quickstart && exasol-quickstart   # with uv

Either form detects your operating system, provisions Exasol the appropriate way, and prints the connection endpoints. No flags, no multi-step setup.

What you get

Component	Endpoint	Purpose
Exasol (database)	127.0.0.1:8563 — user sys / password exasol	the Exasol SQL engine
MCP Server	http://127.0.0.1:4896/mcp	connect an LLM / MCP client to the database
JSON Tables	exasol-quickstart json-tables …	ingest JSON and query it as SQL

Web UI: https://127.0.0.1:8443.

Requirements

The only universal prerequisite is Python 3.9+ with pipx:

python -m pip install --user pipx
python -m pipx ensurepath        # then reopen the terminal

Beyond that, exasol-quickstart chooses how Exasol runs based on your platform:

Operating system	How Exasol runs	Docker
Windows	Exasol Nano, in a container	required (no native Windows engine exists)
macOS (Apple Silicon)	Exasol Personal, in a native VM	not required (experimental)
Linux	Exasol Nano, in a container (native install planned)	required for now

The container-based path is fully tested today; the macOS native path is experimental and not yet validated end to end.

Usage

exasol-quickstart                      # full stack: database + MCP + JSON Tables
exasol-quickstart --no-json-tables     # database + MCP only
exasol-quickstart --dry-run            # print the plan without making changes
exasol-quickstart --base <name>        # force a base: nano-docker | personal | nano-native
exasol-quickstart json-tables --help   # run the JSON Tables CLI

Stop and remove the stack:

docker rm -f exasol-quickstart-db exasol-quickstart-mcp exasol-quickstart-json-tables

Ingesting JSON

docker cp data.json exasol-quickstart-json-tables:/workspace/data.json
exasol-quickstart json-tables ingest-and-wrap --input /workspace/data.json --name demo
# then query it:  SELECT * FROM "EJT_DEMO_VIEW"."demo";

On the container path, the first run pulls the exasol/nano and exasol/mcp-server images and builds the JSON Tables image once (it compiles a small Rust engine — a few minutes). Subsequent runs are fast.

How it works

exasol-quickstart is a single front-door command that detects the platform and assembles the stack accordingly:

• With Docker — Exasol Nano (database), the official exasol/mcp-server image, and a JSON Tables sidecar run as containers on a shared network. This path is tested end to end, including ingest.
• Without Docker — on macOS it uses Exasol Personal (a native VM); on Linux, a native Nano install (planned). The add-ons run as isolated host environments.

MCP Server and JSON Tables have incompatible pyexasol requirements, so each runs in isolation — a separate container, or a separate host environment — never a shared Python environment.

Status

0.3.x — the container path (Nano + MCP + JSON Tables) is tested end to end, including ingest. The no-Docker native bases are selected automatically when Docker is absent; the macOS path is experimental and not yet validated.

License

MIT