v2.1.0

Added

• Model aliases preserve compatibility with renamed model IDs in API requests and chat, inherit the target model's access controls, and are managed through config.yaml or the admin config editor. (#19)
• Publish help documentation on GitHub Pages with rendered architecture and database diagrams.
• Add desktop chat client setup instructions, including ChatWise, to the Connect guide.
• New chats show a notice linking to desktop chat client setup; the notice disappears after the first message. (#63)

Changed

• Retire removed model endpoints while preserving their request history; retired endpoints are excluded from discovery, routing, and health checks, and require a fresh health probe when reactivated.
• GET /v1/models and GET /v1/models/{id} no longer count against the per-key API rate limit.
• Hide a group's Models tab from regular members when no models are granted; owners and admins retain access.
• Link user names in group member lists to their profiles for administrators.
• Help examples use a consistent placeholder host and substitute your instance's URL when viewed in the app.
• Clarify where to save OpenCode configuration and how to show only Lumen models with enabled_providers.
• Refresh help screenshots for chat and model details.
• Update application dependencies.

Fixed

• Config-editor saves preserve the configuration version instead of downgrading version 3 files to version 2.
• Config reloads retry failed model syncs with bounded backoff and report success only after the new configuration is applied.
• A failed startup model sync no longer prevents metrics initialization or produces misleading database migration advice.