#69

What was fixed:

• Created useLocalProviders hook that actively checks for running local provider instances
• Automatically enables detected providers in the settings store
• Provides real-time status with loading states

✅ Improved Ollama Provider

Problem: Hardcoded 8000 token limits regardless of actual model capabilities.
Solution:

• Dynamic context window calculation based on model parameter size
• Proper context windows: 32k for 70B+ models, 16k for 30B+ models, 8k for 7B+ models
• Better model labeling with parameter size and context info

✅ Fixed Provider Settings Management

Problem: Type errors in the provider settings update function.
Solution:

• Fixed updateProviderSettings function signature to properly handle settings updates
• Added correct type imports for IProviderSetting
• Ensured local providers can be properly enabled/disabled

✅ Model Selector Integration

Problem: Detected local providers weren't showing up in the UI selectors.
Solution:

• Local provider detection now integrates with the existing provider/model selection system
• When Ollama or LMStudio are detected running, they automatically appear as available options
• Proper fallback handling when local providers go offline