fix: 修复19个Bug (P0-P3) — 持续性调试第7轮发现的问题

P0 (5): crypto/rand session ID, TTS fallback可达性, goroutine defer recover, adminAuth前缀修正
P1 (5): 普通用户密码验证, context传递, priority clamp, 超时重试, 自主思考速率限制
P2 (4): Briefing AI降级, 前端消息类型渲染, Docker Compose补全, PWA 192图标
P3 (5): goroutine错误处理, .gitignore完善, reminder created_at, voice Dockerfile, Go版本更新

backend/ai-core/Dockerfile

@@ -1,5 +1,5 @@
 # ========== 构建阶段 ==========
-FROM golang:1.23-alpine AS builder
+FROM golang:1.26-alpine AS builder
 
 RUN apk add --no-cache git ca-certificates
 

backend/ai-core/internal/background/thinker.go

@@ -220,6 +220,11 @@ func (t *Thinker) TriggerPostChatThink() {
 	t.wg.Add(1)
 	go func() {
 		defer t.wg.Done()
+		defer func() {
+			if r := recover(); r != nil {
+				log.Printf("[后台思考] 对话后触发 panic 恢复: %v", r)
+			}
+		}()
 
 		// 短暂延迟，让对话有个自然的停顿
 		select {
@@ -261,6 +266,11 @@ func (t *Thinker) resetSilenceTimer() {
 	t.wg.Add(1)
 	go func() {
 		defer t.wg.Done()
+		defer func() {
+			if r := recover(); r != nil {
+				log.Printf("[后台思考] 静默定时器 panic 恢复: %v", r)
+			}
+		}()
 
 		select {
 		case <-t.stopCh:
@@ -316,16 +326,30 @@ func (t *Thinker) HasPendingThoughts() bool {
 // performThink 执行一次增强版后台思考（支持工具调用和记忆管理）
 //
 // triggerReason: "post_chat" (对话后) 或 "silence" (静默超时)
+//
+// 防御性速率限制：即使调用方未检查 minThinkGap，performThink 自身也会
+// 强制执行最小间隔，防止并发调用或 bug 导致 LLM 配额被快速消耗。
 func (t *Thinker) performThink(triggerReason string) {
-	ctx, cancel := context.WithTimeout(context.Background(), 120*time.Second)
-	defer cancel()
-
 	t.mu.Lock()
+	gapSinceLast := time.Since(t.lastThinkTime)
+	minGap := t.minThinkGap
+	if minGap <= 0 {
+		minGap = 5 * time.Second // 默认最小间隔 5 秒
+	}
+	if gapSinceLast < minGap {
+		t.mu.Unlock()
+		log.Printf("[后台思考] 距上次思考仅 %v，跳过 (最小间隔=%v, 触发原因=%s)", gapSinceLast.Round(time.Second), minGap, triggerReason)
+		return
+	}
+
 	t.lastThinkTime = time.Now()
 	t.thinkCount++
 	currentCount := t.thinkCount
 	t.mu.Unlock()
 
+	ctx, cancel := context.WithTimeout(context.Background(), 120*time.Second)
+	defer cancel()
+
 	log.Printf("[后台思考] 开始思考周期 (触发原因=%s, 计数=%d)...", triggerReason, currentCount)
 
 	// 1. 加载人格配置
@@ -356,7 +380,7 @@ func (t *Thinker) performThink(triggerReason string) {
 	// 4. 查询 IoT 设备状态（每次都查询，无间隔限制）
 	var deviceSummary string
 	if t.iotClient != nil {
-		devices := t.iotClient.GetDevicesForContext()
+		devices := t.iotClient.GetDevicesForContext(ctx)
 		if len(devices) > 0 {
 			deviceSummary = formatDeviceContext(devices)
 		}
@@ -460,7 +484,14 @@ func (t *Thinker) performThink(triggerReason string) {
 
 	// 9. 从思考结果中提取记忆（异步）
 	if t.memoryExtractor != nil {
-		go t.extractMemoriesFromThinking(finalContent)
+		go func() {
+			defer func() {
+				if r := recover(); r != nil {
+					log.Printf("[后台思考] 提取记忆 panic 恢复: %v", r)
+				}
+			}()
+			t.extractMemoriesFromThinking(finalContent)
+		}()
 	}
 
 	// 10. 周期性记忆维护（每 10 次思考触发一次，而非按时间）
@@ -649,6 +680,11 @@ func (t *Thinker) storeThought(content string, toolCallsJSON string, toolCallCou
 	// 异步持久化到 memory-service
 	if t.memClient != nil {
 		go func() {
+			defer func() {
+				if r := recover(); r != nil {
+					log.Printf("[后台思考] 持久化思考日志 panic 恢复: %v", r)
+				}
+			}()
 			ctx, cancel := context.WithTimeout(context.Background(), 10*time.Second)
 			defer cancel()
 			if err := t.memClient.SaveThinkingLog(ctx, t.adminUserID, content, toolCallsJSON, toolCallCount, len(content)); err != nil {

backend/ai-core/internal/orchestrator/orchestrator.go

@@ -80,6 +80,11 @@ func (o *Orchestrator) ProcessInput(
 
 	go func() {
 		defer close(eventCh)
+		defer func() {
+			if r := recover(); r != nil {
+				log.Printf("[orchestrator] 编排器主循环 panic 恢复: %v", r)
+			}
+		}()
 
 		// 1. 意图分析
 		intent, err := o.intentAnalyzer.Analyze(ctx, params.Message)

backend/ai-core/internal/subsession/iot_provider.go

@@ -15,11 +15,11 @@ import (
 
 // IoTDeviceProvider IoT 设备查询接口
 type IoTDeviceProvider interface {
-	GetAllDevices() ([]tools.IoTDevice, error)
-	GetDevice(id string) (*tools.IoTDevice, error)
+	GetAllDevices(ctx context.Context) ([]tools.IoTDevice, error)
+	GetDevice(ctx context.Context, id string) (*tools.IoTDevice, error)
 	ToggleDevice(id string) error
 	SetDeviceProperty(id string, field string, value interface{}) error
-	GetDevicesForContext() []tools.IoTDevice
+	GetDevicesForContext(ctx context.Context) []tools.IoTDevice
 }
 
 // IoTProvider IoT 控制子会话提供者
@@ -75,7 +75,7 @@ func (p *IoTProvider) CreateContext(ctx context.Context, params CreateContextPar
 	// 获取当前设备状态
 	var deviceStatusText string
 	if p.iotClient != nil {
-		devices := p.iotClient.GetDevicesForContext()
+		devices := p.iotClient.GetDevicesForContext(ctx)
 		if len(devices) > 0 {
 			deviceStatusText = "当前设备状态：\n"
 			for _, d := range devices {
@@ -208,7 +208,7 @@ func (p *IoTProvider) Execute(ctx context.Context, subCtx []model.LLMMessage) (*
 	msgLower := strings.ToLower(userMessage)
 
 	// 尝试获取设备列表进行匹配
-	devices := p.iotClient.GetDevicesForContext()
+	devices := p.iotClient.GetDevicesForContext(ctx)
 
 	for _, dev := range devices {
 		devNameLower := strings.ToLower(dev.Name)

backend/ai-core/internal/subsession/manager.go

@@ -89,6 +89,11 @@ func (m *Manager) Dispatch(
 		wg.Add(1)
 		go func(p Provider) {
 			defer wg.Done()
+			defer func() {
+				if r := recover(); r != nil {
+					log.Printf("[subsession] dispatch goroutine panic 恢复 (type=%s): %v", p.Type(), r)
+				}
+			}()
 
 			result := model.SubSessionResult{Type: p.Type()}
 
@@ -135,6 +140,11 @@ func (m *Manager) Dispatch(
 
 	// 等待所有子会话完成，关闭通道
 	go func() {
+		defer func() {
+			if r := recover(); r != nil {
+				log.Printf("[subsession] wait goroutine panic 恢复: %v", r)
+			}
+		}()
 		wg.Wait()
 		close(resultCh)
 	}()

backend/ai-core/internal/tools/iot_client.go

@@ -2,6 +2,7 @@ package tools
 
 import (
 	"bytes"
+	"context"
 	"encoding/json"
 	"fmt"
 	"io"
@@ -56,7 +57,7 @@ func NewIoTClient(baseURL string) *IoTClient {
 }
 
 // GetAllDevices 获取所有设备列表（带缓存）
-func (c *IoTClient) GetAllDevices() ([]IoTDevice, error) {
+func (c *IoTClient) GetAllDevices(ctx context.Context) ([]IoTDevice, error) {
 	// 检查缓存
 	c.mu.RLock()
 	if c.cache != nil && time.Since(c.cacheTime) < c.cacheTTL {
@@ -68,7 +69,13 @@ func (c *IoTClient) GetAllDevices() ([]IoTDevice, error) {
 	c.mu.RUnlock()
 
 	// 请求 API
-	resp, err := c.client.Get(c.baseURL + "/api/v1/devices")
+	req, err := http.NewRequestWithContext(ctx, "GET", c.baseURL+"/api/v1/devices", nil)
+	if err != nil {
+		log.Printf("[IoT客户端] 创建请求失败: %v", err)
+		return nil, fmt.Errorf("创建请求失败: %w", err)
+	}
+
+	resp, err := c.client.Do(req)
 	if err != nil {
 		log.Printf("[IoT客户端] 请求失败: %v", err)
 		return nil, fmt.Errorf("获取设备列表失败: %w", err)
@@ -97,8 +104,12 @@ func (c *IoTClient) GetAllDevices() ([]IoTDevice, error) {
 }
 
 // GetDevice 获取单个设备详情
-func (c *IoTClient) GetDevice(id string) (*IoTDevice, error) {
-	resp, err := c.client.Get(c.baseURL + "/api/v1/devices/" + id)
+func (c *IoTClient) GetDevice(ctx context.Context, id string) (*IoTDevice, error) {
+	req, err := http.NewRequestWithContext(ctx, "GET", c.baseURL+"/api/v1/devices/"+id, nil)
+	if err != nil {
+		return nil, fmt.Errorf("创建请求失败: %w", err)
+	}
+	resp, err := c.client.Do(req)
 	if err != nil {
 		return nil, fmt.Errorf("获取设备 %s 失败: %w", id, err)
 	}
@@ -195,8 +206,8 @@ func (c *IoTClient) SetDeviceProperty(id string, field string, value interface{}
 }
 
 // GetDevicesForContext 获取设备状态摘要（供上下文注入使用，失败不报错）
-func (c *IoTClient) GetDevicesForContext() []IoTDevice {
-	devices, err := c.GetAllDevices()
+func (c *IoTClient) GetDevicesForContext(ctx context.Context) []IoTDevice {
+	devices, err := c.GetAllDevices(ctx)
 	if err != nil {
 		log.Printf("[IoT客户端] 获取设备状态摘要失败: %v", err)
 		return nil

backend/ai-core/internal/tools/iot_control_tool.go

@@ -149,22 +149,22 @@ func (t *IoTControlTool) Execute(ctx context.Context, arguments map[string]inter
 
 	// 先获取设备名用于友好的返回消息（失败不影响后续流程）
 	deviceName := deviceID
-	if dev, err := t.iotClient.GetDevice(deviceID); err == nil {
+	if dev, err := t.iotClient.GetDevice(ctx, deviceID); err == nil {
 		deviceName = dev.Name
 	}
 
 	// 处理属性设置类操作
 	switch action {
 	case "set_temperature":
-		return t.handleSetTemperature(deviceID, arguments)
+		return t.handleSetTemperature(ctx, deviceID, arguments)
 	case "set_brightness":
-		return t.handleSetBrightness(deviceID, arguments)
+		return t.handleSetBrightness(ctx, deviceID, arguments)
 	case "set_position":
-		return t.handleSetPosition(deviceID, arguments)
+		return t.handleSetPosition(ctx, deviceID, arguments)
 	case "set_mode":
-		return t.handleSetMode(deviceID, arguments)
+		return t.handleSetMode(ctx, deviceID, arguments)
 	case "set_color":
-		return t.handleSetColor(deviceID, arguments)
+		return t.handleSetColor(ctx, deviceID, arguments)
 	case "turn_off":
 		// 声明式关闭：使用 SetDeviceProperty status/off 而非 toggle
 		// 即使设备已经关闭，SetProperty 也会幂等处理
@@ -204,7 +204,7 @@ func (t *IoTControlTool) Execute(ctx context.Context, arguments map[string]inter
 		}
 
 		// 获取切换后的状态
-		updatedDevice, err := t.iotClient.GetDevice(deviceID)
+		updatedDevice, err := t.iotClient.GetDevice(ctx, deviceID)
 		if err != nil {
 			return &ToolResult{
 				ToolName: "iot_control",
@@ -230,7 +230,7 @@ func extractValue(arguments map[string]interface{}) interface{} {
 }
 
 // handleSetTemperature 处理设置温度
-func (t *IoTControlTool) handleSetTemperature(deviceID string, arguments map[string]interface{}) (*ToolResult, error) {
+func (t *IoTControlTool) handleSetTemperature(ctx context.Context, deviceID string, arguments map[string]interface{}) (*ToolResult, error) {
 	val := extractValue(arguments)
 	if val == nil {
 		return &ToolResult{
@@ -241,7 +241,7 @@ func (t *IoTControlTool) handleSetTemperature(deviceID string, arguments map[str
 	}
 
 	// 先获取当前设备信息
-	currentDevice, err := t.iotClient.GetDevice(deviceID)
+	currentDevice, err := t.iotClient.GetDevice(ctx, deviceID)
 	if err != nil {
 		return &ToolResult{
 			ToolName: "iot_control",
@@ -275,7 +275,7 @@ func (t *IoTControlTool) handleSetTemperature(deviceID string, arguments map[str
 }
 
 // handleSetBrightness 处理设置亮度
-func (t *IoTControlTool) handleSetBrightness(deviceID string, arguments map[string]interface{}) (*ToolResult, error) {
+func (t *IoTControlTool) handleSetBrightness(ctx context.Context, deviceID string, arguments map[string]interface{}) (*ToolResult, error) {
 	val := extractValue(arguments)
 	if val == nil {
 		return &ToolResult{
@@ -286,7 +286,7 @@ func (t *IoTControlTool) handleSetBrightness(deviceID string, arguments map[stri
 	}
 
 	// 先获取当前设备信息
-	currentDevice, err := t.iotClient.GetDevice(deviceID)
+	currentDevice, err := t.iotClient.GetDevice(ctx, deviceID)
 	if err != nil {
 		return &ToolResult{
 			ToolName: "iot_control",
@@ -320,7 +320,7 @@ func (t *IoTControlTool) handleSetBrightness(deviceID string, arguments map[stri
 }
 
 // handleSetPosition 处理设置窗帘位置
-func (t *IoTControlTool) handleSetPosition(deviceID string, arguments map[string]interface{}) (*ToolResult, error) {
+func (t *IoTControlTool) handleSetPosition(ctx context.Context, deviceID string, arguments map[string]interface{}) (*ToolResult, error) {
 	val := extractValue(arguments)
 	if val == nil {
 		return &ToolResult{
@@ -330,7 +330,7 @@ func (t *IoTControlTool) handleSetPosition(deviceID string, arguments map[string
 		}, nil
 	}
 
-	currentDevice, err := t.iotClient.GetDevice(deviceID)
+	currentDevice, err := t.iotClient.GetDevice(ctx, deviceID)
 	if err != nil {
 		return &ToolResult{
 			ToolName: "iot_control",
@@ -364,7 +364,7 @@ func (t *IoTControlTool) handleSetPosition(deviceID string, arguments map[string
 }
 
 // handleSetMode 处理设置空调模式
-func (t *IoTControlTool) handleSetMode(deviceID string, arguments map[string]interface{}) (*ToolResult, error) {
+func (t *IoTControlTool) handleSetMode(ctx context.Context, deviceID string, arguments map[string]interface{}) (*ToolResult, error) {
 	val := extractValue(arguments)
 	if val == nil {
 		return &ToolResult{
@@ -383,7 +383,7 @@ func (t *IoTControlTool) handleSetMode(deviceID string, arguments map[string]int
 		}, nil
 	}
 
-	currentDevice, err := t.iotClient.GetDevice(deviceID)
+	currentDevice, err := t.iotClient.GetDevice(ctx, deviceID)
 	if err != nil {
 		return &ToolResult{
 			ToolName: "iot_control",
@@ -408,7 +408,7 @@ func (t *IoTControlTool) handleSetMode(deviceID string, arguments map[string]int
 }
 
 // handleSetColor 处理设置灯光颜色
-func (t *IoTControlTool) handleSetColor(deviceID string, arguments map[string]interface{}) (*ToolResult, error) {
+func (t *IoTControlTool) handleSetColor(ctx context.Context, deviceID string, arguments map[string]interface{}) (*ToolResult, error) {
 	val := extractValue(arguments)
 	if val == nil {
 		return &ToolResult{
@@ -427,7 +427,7 @@ func (t *IoTControlTool) handleSetColor(deviceID string, arguments map[string]in
 		}, nil
 	}
 
-	currentDevice, err := t.iotClient.GetDevice(deviceID)
+	currentDevice, err := t.iotClient.GetDevice(ctx, deviceID)
 	if err != nil {
 		return &ToolResult{
 			ToolName: "iot_control",

backend/ai-core/internal/tools/iot_tools.go

@@ -47,7 +47,7 @@ func (t *IoTQueryTool) Execute(ctx context.Context, arguments map[string]interfa
 
 	if deviceID != "" {
 		// 查询单个设备
-		device, err := t.iotClient.GetDevice(deviceID)
+		device, err := t.iotClient.GetDevice(ctx, deviceID)
 		if err != nil {
 			return &ToolResult{
 				ToolName: "iot_query",
@@ -63,7 +63,7 @@ func (t *IoTQueryTool) Execute(ctx context.Context, arguments map[string]interfa
 	}
 
 	// 查询所有设备
-	devices, err := t.iotClient.GetAllDevices()
+	devices, err := t.iotClient.GetAllDevices(ctx)
 	if err != nil {
 		return &ToolResult{
 			ToolName: "iot_query",

backend/ai-core/internal/tools/tool_engine_client.go

@@ -8,6 +8,7 @@ import (
 	"io"
 	"log"
 	"net/http"
+	"strings"
 	"time"
 )
 
@@ -82,7 +83,54 @@ func (c *ToolEngineClient) GetDefinitions(ctx context.Context) ([]ToolDefinition
 }
 
 // Execute 通过 tool-engine 执行工具调用
+// 包含重试逻辑：最多重试 2 次（共 3 次尝试），间隔 100ms
 func (c *ToolEngineClient) Execute(ctx context.Context, toolName string, arguments map[string]interface{}) (*ToolResult, error) {
+	const maxRetries = 2
+	const retryDelay = 100 * time.Millisecond
+
+	var lastErr error
+	for attempt := 0; attempt <= maxRetries; attempt++ {
+		if attempt > 0 {
+			log.Printf("[tool-engine-client] 工具 %s 第 %d 次重试 (上次错误: %v)", toolName, attempt, lastErr)
+			select {
+			case <-ctx.Done():
+				return &ToolResult{
+					ToolName: toolName,
+					Success:  false,
+					Error:    fmt.Sprintf("请求被取消: %v", ctx.Err()),
+				}, nil
+			case <-time.After(retryDelay):
+			}
+		}
+
+		result, err := c.executeOnce(ctx, toolName, arguments)
+		if err == nil && result.Success {
+			return result, nil
+		}
+		if result != nil {
+			lastErr = fmt.Errorf("%s", result.Error)
+		} else {
+			lastErr = err
+		}
+
+		// 不可重试的错误：工具不存在、参数序列化失败、创建请求失败
+		if result != nil && (strings.Contains(result.Error, "不存在") ||
+			strings.Contains(result.Error, "序列化") ||
+			strings.Contains(result.Error, "创建请求")) {
+			return result, nil
+		}
+	}
+
+	log.Printf("[tool-engine-client] 工具 %s 所有重试均失败 (最后错误: %v)", toolName, lastErr)
+	return &ToolResult{
+		ToolName: toolName,
+		Success:  false,
+		Error:    fmt.Sprintf("请求 tool-engine 失败 (已重试 %d 次): %v", maxRetries, lastErr),
+	}, nil
+}
+
+// executeOnce 执行单次工具调用（不含重试逻辑）
+func (c *ToolEngineClient) executeOnce(ctx context.Context, toolName string, arguments map[string]interface{}) (*ToolResult, error) {
 	body, err := json.Marshal(map[string]interface{}{
 		"arguments": arguments,
 	})