v0.1.1

Changelog

0.1.1 - 2026-07-17

Fixed

• Show model request failures directly in the conversation instead of silently rendering an empty assistant reply. Provider error details such as an invalid API key response are preserved, with an actionable fallback when the provider returns no detail.
• Keep failed assistant messages visible after conversation grouping and reload while continuing to hide harmless empty responses.
• Add regression coverage for provider failures, missing error details, and normal empty assistant messages.