Skip to content

quickstart_windows.md

Latest commit

 

History

History
102 lines (67 loc) · 2.05 KB

quickstart_windows.md

File metadata and controls

102 lines (67 loc) · 2.05 KB
Raw

Windows Quick Start

This is the fastest path to a working LumaKit install on Windows.

1. Install Ollama

Install Ollama first and make sure it is running locally.

In PowerShell:

ollama list

If that works, LumaKit can talk to your local Ollama endpoint.

2. Pick a model

Choose the model you want LumaKit to use and make sure your Ollama setup can serve it.

If you want the strongest first-run impression, use a strong tool-capable model. Small models can chat, but they are more brittle with multi-step tool loops.

3. Clone the repo

git clone https://github.com/patmakesapps/LumaKit.git
cd LumaKit

4. Install dependencies

py -m pip install -r requirements.txt
playwright install chromium
py -m pip install -e .

5. Create .env

copy .env.example .env

Set at least:

OLLAMA_MODEL="your-model-here"

Optional but common:

OLLAMA_FALLBACK_MODEL="your-fallback-model"
LUMAKIT_WEB_PORT="7865"

If you prefer, you can leave OLLAMA_MODEL blank and choose your primary model from the web UI on first launch. The app will block chat until a model is selected.

6. Launch LumaKit

lumakit open

That should:

  • start the backend if it is not already running
  • reuse it if it is already running
  • open the web UI in your browser

If no model is configured in .env or app settings yet, LumaKit opens into a first-run setup state and asks you to choose one in Settings before chatting.

7. Install the Windows shortcuts

lumakit shortcut install

When native shortcut creation succeeds, that writes:

  • a Desktop shortcut
  • a Start Menu shortcut

Both launch lumakit open.

Useful commands

lumakit status
lumakit stop
lumakit serve
  • lumakit status shows whether the backend is already running
  • lumakit stop stops the backend
  • lumakit serve runs in the foreground for debugging

Optional next steps