[ENHANCEMENT] Add debug thoughts display for LM Studio to detect infinite loops in OSS models #8140
Description
Problem (one or two sentences)
Some open-source models (like oss-gpt-20b) can enter infinite reasoning loops when using LM Studio provider, but users have no visibility into the model's thought process to detect when this occurs.
Context (who is affected and when)
LM Studio users working with local open-source models, particularly those experimenting with reasoning-capable models like oss-gpt-20b. This happens during normal chat interactions when the model gets stuck in internal reasoning cycles.
Desired behavior (conceptual, not technical)
Users should be able to see the model's internal reasoning/thinking process in real-time to identify when a model is stuck in a loop and understand what the model is trying to process.
Constraints / preferences (optional)
Should be optional/toggleable to avoid UI clutter for users who don't need it
Real-time display without significant performance impact
Clear distinction between regular output and debug thoughts
Accessible through settings panel
Request checklist
- I've searched existing Issues and Discussions for duplicates
- This describes a specific problem with clear context and impact
Roo Code Task Links (optional)
No response
Acceptance criteria (optional)
Given a user has LM Studio configured with a reasoning-capable model
When the user enables "Show Debug Thoughts" option in LM Studio settings
Then the model's internal reasoning should be displayed in real-time during responses
And users can easily distinguish between regular output and debug thoughts
And the feature can be toggled on/off without requiring restart
But it should not impact performance when disabled
Proposed approach (optional)
Add a "Show Debug Thoughts" checkbox in LM Studio provider settings that enables display of reasoning output from models that support it. The reasoning text would be shown with different styling to distinguish it from regular model responses.
Trade-offs / risks (optional)
May increase token usage display complexity
Could overwhelm users with too much information if not properly styled
Performance consideration when processing high-frequency reasoning output
Alternative: Could implement as a separate debug panel instead of inline display