v0.0.9

v0.0.9

New Features

• Multimodal Messages: Full multimodal support across all channels — Lark/WeChat/WeCom/Web can send and receive images; Web UI supports drag-drop and paste image input with lightbox preview; agents process image content from MCP tool results
• WeChat Channel: New channel.wechat package with full WeChat integration — API client, QR code login, file sending/receiving
• Mobile Responsive Web UI: All web dashboard views now support mobile display
• Global MCP & Skills: Support for global MCP skills configuration and agent-level MCP/Skill management
• Shared Conversation History: Agents can now share conversation history across sessions
• Auto Start: Server auto-start capability with startup configuration

Architecture Changes

• Multimodal Content Pipeline: MessageContent supports both string and multipart array formats; extracted shared MessageList component and messageRender.ts utility for unified rendering of text, images, tool calls, and streaming
• WebSocket Refactoring: Reorganized WebSocketUserService, unified event types, expanded createProcessAIHandler processing pipeline
• SkillService No-Cache: Switched to no-cache mode for real-time skill updates
• Code Cleanup: Removed unused CommandDecorators, simplified redundant chat components

Improvements

• Channel File & Image Support: WeChat/WeCom support file sending/receiving and mixed image-text content
• Channel Password Auth: Channel management UI now supports password/auth configuration
• Various Bug Fixes: Fixes across Lark/Slack session handlers, message serializer, model config, and embedding types