This repository contains the necessary files needed to self-host the Server Witch API to connect the Server Witch client to the Server Witch custom GPT. You can find more information about the Server Witch client here.
This server is built in python with Quart and Hypercorn, and can be deployed with docker.
You can either deploy the API only, or use the provided docker compose file which includes an nginx frontend.
To build the docker image, run:
docker build -t serverwitch-api .
Then, run it with:
docker run -p 8000:8000 --name serverwitch-api serverwitch-api:latest
The server will listen on port 8000 (HTTP). You need to setup your own reverse proxy to support TLS.
For your convenience, a docker compose file is provided to deploy the Server Witch API behind an nginx proxy, using certbot to obtain TLS certificates. To use it, run:
DOMAIN_NAME="example.com" EMAIL_ADDRESS="email@example.com" docker compose up --build
Replace example.com and email@example.com with the domain name and email address you wish to use for the TSL certificate.
If you deploy your own Server Witch API, you will also need your own custom GPT that uses your API server. You can make it using the official GPT editor. You can use the provided prompt and OpenAPI specification for the Actions. In the OpenAPI file, replace the url field with your own domain name.
Note that it is also possible to use any other LLM that can send requests based on the OpenAPI specification.
This project is built with poetry:
poetry build
To launch the server in debug mode (without Hypercorn):
poetry install
poetry run start
Contributions are highly appreciated. Please feel free to submit pull requests and bug reports. If you discover a security vulnerability in this software, contact me at g33kex[at]pm[dot]me.
If you want to discuss with us, we're also on matrix #serverwitch:beyond-undefined.fr