claude-code - 💡(How to fix) Fix [Feature Request] Add fallback prompt for auto-mode classifier soft-denials instead of returning denial to model [1 comments, 2 participants]

TL;DR

Implementing a permissions.autoModeFallback setting to interrupt and prompt the user on classifier soft-denies could address the issue of subpar results due to auto-mode workarounds.

Guidance

• Introduce a new setting, permissions.autoModeFallback, with options like "ask" to handle classifier soft-denies by interrupting and prompting the user.
• Evaluate the effectiveness of this setting in preventing subpar results and silent skipping of steps.
• Consider the trade-offs between the ergonomic benefits of auto mode and the need for explicit user confirmation in certain cases.
• Test this setting with various scenarios to ensure it triggers the interactive permission prompt as expected.

Example

No code snippet is provided as the issue does not contain specific code references.

Notes

The proposed solution relies on the introduction of a new setting, which may require updates to the existing permissions system. The effectiveness of this solution depends on the specific implementation and user interaction.

Recommendation

Apply workaround: Implement the proposed permissions.autoModeFallback setting to address the issue of subpar results due to auto-mode workarounds, as it provides a more immediate and user-centric solution.