Skip to content

The new BARC study is here

Download now
Image
Try for free
Contact us Try for free
Image

Home » Developers

Local Agent-Ready Starter Kit

The Agentic Database.

Built for autonomous agents to go from idea to production: fast, unassisted, without burning tokens.

One command installs the full local agentic stack on your own machine: the database, AI access, and drivers.

curl https://www.exasol.com/install/starter-kit.sh | sh

macOS runs the database natively. Linux and WSL use Podman. Run your first AI-assisted query in under two minutes.

This installs the full stack, not just the database

Exasol database

exapump

MCP server

pyexasol 

irm https://www.exasol.com/install/starter-kit.ps1 | iex

Run from PowerShell. Requires Podman. Same four components installed as macOS/Linux.

This installs the full stack, not just the database

Exasol database

exapump

MCP server

pyexasol

See it in action

One command. One trusted answer.

Watch the whole flow: install, connect an AI assistant, ask a real question, see the SQL before it runs, and get an answer you can verify yourself.

  1. Install
    One commands sets up everything you need.
  2. Load data
    Sample data ready instantly, or point it at your own.
  3. Connect your agent
    Claude, Cursor, or Codex, connected over MCP in one step.
  4. Query
    Ask a real question. See the SQL. Get an answer you trust.

Why agents want local first

What changes when your user
is an agent, not a person.

Cloud services assume a human is standing by to log in, approve budget, and grant access. Agents don’t have that luxury, so the fastest path for them is the one that doesn’t ask.

No login required

Cloud services need an account an agent can’t create on its own. Local needs nothing.

No cost anxiety

Agentic workload is unpredictable ahead of time. A local instance can’t run up a surprise bill.

No permission needed

Spinning up cloud workloads usually means asking someone first. Local sidesteps that entirely.

Zero production risk

Isolated by default. Nothing an agent does locally can touch a workload that matters.

Second-level iteration

Agents want to spin up, test, and tear down constantly. Local round-trips in seconds, not minutes.
Image

What it means to be agentic

An agentic database lets an autonomous agent build a solution to your problem: fast, unassisted, and without burning through its token budget.

Most data platforms try to do everything so you never have to leave. A starter kit works differently. It is a small, focused core plus a few essential pieces: a database, a way to load your data, a way to script against it, and a way for an AI agent to query it safely. Pick what you need. Nothing extra to learn first.

Explore the ecosystem

Not a platform. A toolbox.

Not a platform to adopt wholesale. It is a set of components built to be picked up individually, the way an agent would reach for exactly what a problem needs.

Exasol
exasol-personal

The Analytics Database for Agentic AI, free for personal use. Deploy full-scale Exasol to your cloud or Mac in minutes.

Go View on GitHub
Exasol
mcp-server

The official MCP Server. How AI agents discover schema and query Exasol safely, and the core of the agentic access layer.

Python View on GitHub
Exasol-Labs
exasol-agent-skills

Exasol skills for AI agents, optimized for Claude Code and OpenAI Codex. Teach your coding agent how to work with Exasol.

Markdown View on GitHub
Exasol
pyexasol

The official Python driver, WebSocket-native and ODBC-free. Parallel HTTP transport delivers fast pandas and Polars dataframes.

Python View on GitHub
Exasol-Labs
exarrow-rs

ADBC-compatible, Arrow-native driver written in Rust. Zero-copy query results for Polars, Python, and the whole Arrow ecosystem.

Rust View on GitHub
Exasol
dbt-exasol

The official dbt adapter for Exasol. Model, test, and document your data transformations right inside the database.

Python View on GitHub
Exasol-Labs
exasol-vscode

A full SQL IDE inside VS Code. Connection management, intelligent SQL editing, query execution, and schema exploration.

TypeScript View on GitHub
Exasol
powerbi-exasol

The Microsoft-certified Power BI connector, built into Power BI Desktop. Live DirectQuery dashboards straight on Exasol.

Power Query M View on GitHub
Exasol-Labs
exasol-json-tables

Ingest, query, and reshape nested JSON with JSON-native SQL. Dotted paths, array joins, relational speed. Experimental.

Python View on GitHub
Explore Exasol on GitHub
Explore Exasol Labs on GitHub

One database, all the way up

Start local. Scale without switching engines.

The path an agent takes locally is the same path a production deployment takes later: same SQL, same drivers, same MCP config.

Start here

Local

One command. No account. Easy to spin up.

Zero setup friction

Scale up

Personal · Cloud

Same engine, deployed to AWS, Azure, Exoscale, or STACKIT.

Same SQL

Production

Enterprise

On-prem, hybrid, or multi-cloud, with zero rewrite from where you started.

Zero rewrite

Ask

Inspect

Run

Validate

Rerun

Start locally. Connect AI safely.
Inspect the SQL. Rerun the workflow.

No cloud account, no admin rights, no sales call. One command, and your own AI-ready analytics stack is running.

Get the install command
View source on GitHub
Read the full quickstart