feat(ai-core): trace instrumentation, LLM thinker, decay scheduler

- orchestrator: Tracer interface with spans for intent analysis, persona,
  subsession dispatch, context build, LLM streaming and each tool call;
  traces finish when the stream completes — the admin Trace waterfall
  finally receives data
- background: Thinker generates real LLM thoughts (post-chat + silence
  triggers, recent history as context, empty/overlong outputs discarded);
  thoughts broadcast to web clients as 'thinking' WS messages; pipeline
  postChatHook triggers post-chat thinking after each round
- memory: ApplyDecayAll + ListUserIDs with a 6h decay/purge cron in main
  (was implemented but never scheduled)

backend/cmd/cyrene/main.go

@@ -32,6 +32,7 @@ import (
 	webfetch "git.yeij.top/AskaEth/Cyrene_v2/plugins/web-fetch"
 	websearch "git.yeij.top/AskaEth/Cyrene_v2/plugins/web-search"
 
+	"git.yeij.top/AskaEth/Cyrene_v2/services/ai-core/background"
 	"git.yeij.top/AskaEth/Cyrene_v2/services/ai-core/llm"
 	llmlog "git.yeij.top/AskaEth/Cyrene_v2/services/ai-core/llm"
 	tracepkg "git.yeij.top/AskaEth/Cyrene_v2/services/ai-core/trace"
@@ -244,6 +245,37 @@ func main() {
 	perfMon := perfpkg.NewMonitor(120, 10*time.Second)
 	perfMon.Start()
 
+	// Wire tracing into the dialogue pipeline
+	orch.SetTracer(pipelineTracer{traceLogger})
+
+	// === 12.5 Background Thinker ===
+	thinker := background.New(true, 5*time.Minute, 30*time.Minute)
+	thinker.SetLLM(&memory.LLMAdapterBridge{Adapter: llmSelector.DefaultAdapter()})
+	thinker.SetHistoryProvider(func(sessionID string, limit int) []model.Message {
+		msgs := ctxStore.Get(sessionID)
+		if len(msgs) > limit {
+			msgs = msgs[len(msgs)-limit:]
+		}
+		return msgs
+	})
+	thinker.Start(context.Background())
+	orch.SetPostChatHook(thinker.TriggerPostChat)
+
+	// === 12.6 Memory decay scheduler ===
+	if memSvc != nil {
+		go func() {
+			ticker := time.NewTicker(6 * time.Hour)
+			defer ticker.Stop()
+			for range ticker.C {
+				if n, err := memSvc.ApplyDecayAll(context.Background()); err != nil {
+					logger.Infof("[memory] decay run failed: %v", err)
+				} else if n > 0 {
+					logger.Infof("[memory] decay applied (%d memories updated)", n)
+				}
+			}
+		}()
+	}
+
 	gwDeps := api.Dependencies{
 		AuthMgr:      authMgr,
 		Orchestrator: orch,
@@ -267,6 +299,13 @@ func main() {
 	wsHub := ws.NewHub(orch)
 	gwDeps.WSHub = wsHub
 
+	// Push background thoughts to connected web clients
+	go func() {
+		for th := range thinker.ThoughtChan() {
+			wsHub.Broadcast(map[string]any{"type": "thinking", "content": th.Content})
+		}
+	}()
+
 	apiRouter := api.New(gwDeps)
 
 	wsMux := http.NewServeMux()

backend/cmd/cyrene/trace_adapter.go

@@ -40,6 +40,12 @@ func (a *traceAdapter) Recent(n int) []map[string]any {
 
 func (a *traceAdapter) Clear() { a.Logger.Clear() }
 
+// pipelineTracer adapts trace.Logger to orchestrator.Tracer
+// (StartTrace's return value is discarded at the call site).
+type pipelineTracer struct{ *tracepkg.Logger }
+
+func (p pipelineTracer) StartTrace(id, userMsg string) { p.Logger.StartTrace(id, userMsg) }
+
 func spanToMap(s tracepkg.Span) map[string]any {
 	m := map[string]any{
 		"name": s.Name, "label": s.Label,