ranade-oss fixes the front door to multi-provider AI
A pair of UI fixes makes sure firms can actually see and enter the credentials needed to use their approved AI providers.
The fork had a frustrating failure mode: key settings for major AI providers could disappear when the provider catalogue had not fully loaded. A separate mismatch also made one available OpenAI model appear unavailable.
ranade-oss fixed that gating problem in two parts of the settings experience, then added regression coverage. The practical result is more reliable setup for organisations using several model providers, rather than an admin being told a service is missing when the issue is only the interface.
The follow-up matters because it closes the same loophole in a second view, suggesting the team is checking the whole setup journey rather than treating the first report as isolated.
Spotted something wrong? Or know the PR text has fresher detail than the writeup above?