v1.1.0: Live Models, Custom Claude Icon & Enhanced Security

What's Changed in v1.1.0

• Live Model Discovery & Badges: Wired settings model grid to live Anthropic /v1/models\ API instead of static list (props @niklasbaldi). Cards now display context window pills (e.g. \200k, \1M) and \Thinking\ badges for hybrid reasoning models.
• Claude Icon in Settings: Replaced the generic gear wheel icon in DSH Settings navigation with the authentic Anthropic Claude icon badge, and upgraded the settings panel header.
• Least-Privilege Security Scopes: Permanently removed the high-risk \org:create_api_key\ scope. Default scopes now strictly require \user:profile user:inference\ (props @kyale), with configurable \process.env.CLAUDE_OAUTH_SCOPES\ override.
• HTML Sanitization on Callback Server: Escaped OAuth callback error and profile values to prevent reflected injection on local callback ports.
• Multi-Account Roadmap: Documented and prepared architecture for multi-account switching (#3).
• Test Suite Expansion: Added unit tests for HTML escaping and scope least-privilege enforcement (all 7 tests passing).

v1.0.0: Initial Release

First official open-source release of dsh-claude-oauth. Adds Google/Gmail Claude OAuth sign-in to DeepSeek Harness Settings, live quota tracking, and automatic token synchronization.