Bug report
Summary
After a fresh user configures their first AI provider, the main chat composer can continue to show Set up AI provider instead of refreshing to the newly available model selector.
Steps to reproduce
- Start OpenBridge with no configured AI providers.
- Open Settings > AI Providers.
- Configure the first provider, for example by signing in with OpenAI OAuth or saving an API token.
- Return to the main chat window.
Expected behavior
The main chat composer should immediately reload provider/model state and show the configured provider's model selector without requiring an app restart or manually reopening the model menu.
Actual behavior
The main chat composer can remain in the empty provider state and still show Set up AI provider, even though the provider settings and credentials were saved successfully.
Notes
This appears to be stale UI state in ChatEditorViewModel.availableModelGroups: provider settings are persisted, but the already-open composer does not reliably reload model availability after provider configuration changes.
Draft PR #27 appears to address this area as part of a broader runtime settings hot-reload change. A narrow fix would be to broadcast AI provider store mutations and have the chat composer reload selected model/provider state when that notification is received.
Bug report
Summary
After a fresh user configures their first AI provider, the main chat composer can continue to show
Set up AI providerinstead of refreshing to the newly available model selector.Steps to reproduce
Expected behavior
The main chat composer should immediately reload provider/model state and show the configured provider's model selector without requiring an app restart or manually reopening the model menu.
Actual behavior
The main chat composer can remain in the empty provider state and still show
Set up AI provider, even though the provider settings and credentials were saved successfully.Notes
This appears to be stale UI state in
ChatEditorViewModel.availableModelGroups: provider settings are persisted, but the already-open composer does not reliably reload model availability after provider configuration changes.Draft PR #27 appears to address this area as part of a broader runtime settings hot-reload change. A narrow fix would be to broadcast AI provider store mutations and have the chat composer reload selected model/provider state when that notification is received.