A ChatGPT-style interface built for server-side extensibility.
Haven't yet tested setup on a fresh machine. But this probably works.
- docker desktop (or anything that gives you the docker compose CLI)
- An LLM API key
Gemini has a free tier, though I don't recommend it for anything beyond quick testing.
git clone git@github.com:utkashd/k4.git
cd k4use `smake` (recommended)
./smake devThis is just running docker compose up along with some extra conveniences
(ensures docker desktop is running, removes any old dangling images, prints some
useful messages). You are encouraged to read exactly what this does before
running it: https://github.com/utkashd/k4/blob/main/smake
Now you can navigate to http://localhost:5173. The server will be http://localhost:8000, and if you care, the API docs are at http://localhost:8000/docs. Setup is now done.
When you are done, you can stop K4 with
./smake stopwhich is basically just running docker compose stop.
OR use `docker compose up`
K4_ENVIRONMENT=development docker compose up --build -dand stop K4 with
docker compose down