4 Commits

Author	SHA1	Message	Date
ZhenYi	c6bb72682b	fix(agent): 修复扣费链路并实现级联扣费策略 ... - billing.rs: 修复参数传递 (model_id -> version_id) - billing.rs: 新增 BillingResult 枚举支持 InsufficientBalance 错误 - billing.rs: 实现级联扣费 (优先 project 余额，不足时 fallback 到 workspace) - billing.rs: 余额不足时创建系统消息并持久化 - chat/service.rs: 捕获 InsufficientBalance 错误并调用 create_system_message - client/mod.rs: 超时时间从 60s 改为 120s	2026-04-28 19:59:06 +08:00
ZhenYi	5b3a6700be	refactor(agent): replace custom ReAct loop with rig::agent::Agent ... - Use AgentBuilder for native tool-calling with stream_prompt() - Add RecordingTool wrapper preserving retry + DB recording - Fix tool_choice bug in do_completion (same as call_stream_once) - Add seq field to RoomMessageStreamChunkEvent for strict ordering - Map streaming events: Text→Answer, Reasoning→Thought, ToolCall→Action - Only final event has done=true, removed premature stream ending - Store __chunks__ JSON in thinking_content for ordered replay	2026-04-28 09:42:36 +08:00
ZhenYi	b4b5538447	feat(agent): add ordered stream chunk collection + retry for HTTP errors ... - StreamChunk/StreamChunkType types for preserving arrival order - Chunk collection in call_stream_once and process_stream - Add "error sending request" and "Http client error" to retryable errors - StreamResult includes chunks vector for ordered replay	2026-04-26 13:10:26 +08:00
ZhenYi	10c0cc007b	refactor(agent): split into submodules and add Qdrant embedding ... - Split agent crate into client/, model/, agent/ subdirs - Add billing.rs for token usage recording - Add sync.rs for upstream model sync - EmbedService: Qdrant-backed vector memory for semantic search - ChatService: wire EmbedService for memory lookup, passive skill awareness - ReAct loop: streamline with tokio::select! and proper error handling	2026-04-25 20:09:33 +08:00