nipunbatra8 bets the whole stack on OpenAI
This fork tears out its Claude and Gemini support and standardizes on a single AI provider - OpenAI - for everything.
nipunbatra8 made a clean directional call: instead of letting the app talk to several AI vendors, it now speaks only to OpenAI. The old multi-provider plumbing - and the user-facing controls for picking between models - are gone, replaced by a short menu of current OpenAI models and a single API key for the whole server.
It's a commitment, not an experiment, touching twenty-odd files. Two things to watch if you're evaluating it: the work to let each user bring their own key looks half-finished, so every request currently runs on one shared key, and one of the higher-end reasoning models on offer appears misconfigured and may fail at runtime. Worth a look on GitHub if you're weighing whether to consolidate on one vendor - and a caution if you're not.
Spotted something wrong? Or know the PR text has fresher detail than the writeup above?