Getting Started with Kontext Labs

Kontext Labs runs entirely on your machine. It bundles a local FastAPI server, an embedded SQLite database, and the Kontext web UI into a single Python package — no cloud account required to get started.

Prerequisites

Python 3.12 or newer
One of the following installers (recommended):
- pipx — installs CLI tools into isolated environments
- uv — fast Python package manager with built-in tool support
A running LLM provider (local or remote). Ollama is recommended for fully offline use.

Install

Choose either installer:

# pipx (recommended for end-users)
pipx install kontext-labs

# uv (recommended for developers)
uv tool install kontext-labs

Both install kontext-labs and kontext-labs-serve into your PATH.

Launch

kontext-labs serve

On first run, a short setup wizard creates your local profile at ~/.kontext/. After setup, the server starts and the Kontext UI is available at http://127.0.0.1:8000.

To open the browser automatically:

kontext-labs serve --open

Configure an LLM Provider

Kontext Labs does not bundle an LLM — bring your own. After launching, open the UI and go to LLM Providers to connect a provider.

Supported options include:

Ollama (local, offline) — run ollama serve before starting Kontext Labs
OpenAI, Anthropic, Azure OpenAI — add your API key in the provider settings
Any OpenAI-compatible endpoint

See LLM Providers for the full list of supported runtimes and model IDs.

CLI Commands

kontext-labs is a full CLI beyond just serve:

Command	Description
`kontext-labs serve`	Start the local server and UI
`kontext-labs project list`	List local projects
`kontext-labs project create <name>`	Create a new project
`kontext-labs project import <files…>`	Import data files into a project
`kontext-labs query "<sql>"`	Run a SQL query against a local project
`kontext-labs job list`	List available job types
`kontext-labs job run <type>`	Run a job synchronously

Supported Platforms

Platform	Architecture
macOS	arm64, x86_64
Linux	x86_64 (manylinux)
Windows	x86_64

Python 3.12 is the only supported Python version for the first release.

Next Steps

LLM Providers — connect and configure your language model
Data Catalogs — connect external databases and file sources
Notebooks — run SQL, Python, and AI workflows in the browser
Getting Started with Kontext Labs Platform — cloud-hosted version of Kontext