[0.5.1] - 2026-03-10
Added
- Added queued message lifecycle controls: queue pause/status hints, queue count/preview metadata, dedicated confirm flow for clearing queued drafts, and localized queue messages in English/Chinese.
- Added service-tier support to request configuration so compatible providers can send
service_tierheaders (priority) when fast mode is enabled, with model-levelserviceTiersettings in providers. - Added User-Agent propagation in provider model discovery requests (including OpenAI provider routes), and set a dedicated User-Agent for Qwen provider defaults.
Changed
- Enhanced queueing behavior in chat actions: stopping an active run now pauses queue dispatch, pending messages continue in order after resume/dispatch state recovery, and queued items are cleared together when session messages are cleared.
- Improved AskUserQuestion flow with modular question block rendering, previous/next/submit navigation, and stronger state-driven answer capture.
- Refined the Assistant message detail dialog request-body section and made plan execution flow more robust by opening the right panel automatically when implementing the active session plan.
- Cleaned chat model popover wiring by removing an unused provider prop from
ModelSwitcherusage.