fix(context): address review - clarify threshold alignment, i18n strings, add test coverage

pkg/agent/context_usage.go

@@ -63,6 +63,13 @@ func computeContextUsage(agent *AgentInstance, sessionKey string) *bus.ContextUs
 
 	// summarizeAt = soft summarization trigger: matches maybeSummarize's
 	// threshold (contextWindow * SummarizeTokenPercent / 100).
+	//
+	// NOTE: The engine's maybeSummarize compares this threshold against
+	// history-message tokens only (via estimateTokens), while UsedTokens
+	// (shown in /context) includes system prompt, summary, and tool
+	// definitions on top of history tokens.  A "Used > SummarizeAt" display
+	// does not necessarily mean summarization will fire — the engine may
+	// still consider the history-token budget to be under threshold.
 	summarizeAt := contextWindow * agent.SummarizeTokenPercent / 100
 	if summarizeAt <= 0 {
 		summarizeAt = compressAt

pkg/channels/pico/pico_test.go

@@ -602,10 +602,11 @@ func TestBeginStream_FinalizeIncludesContextUsage(t *testing.T) {
 		t.Fatal("streamer should support FinalizeWithContext")
 	}
 	if err := contextStreamer.FinalizeWithContext(context.Background(), "final", &bus.ContextUsage{
-		UsedTokens:       10,
-		TotalTokens:      100,
-		CompressAtTokens: 80,
-		UsedPercent:      10,
+		UsedTokens:        10,
+		TotalTokens:       100,
+		CompressAtTokens:  80,
+		SummarizeAtTokens: 60,
+		UsedPercent:       10,
 	}); err != nil {
 		t.Fatalf("FinalizeWithContext() error = %v", err)
 	}
@@ -627,6 +628,9 @@ func TestBeginStream_FinalizeIncludesContextUsage(t *testing.T) {
 	if got := rawUsage["used_tokens"]; got != float64(10) {
 		t.Fatalf("used_tokens = %#v, want 10", got)
 	}
+	if got := rawUsage["summarize_at_tokens"]; got != float64(60) {
+		t.Fatalf("summarize_at_tokens = %#v, want 60", got)
+	}
 }
 
 func TestCreateAndAddConnection_RespectsMaxConnectionsConcurrently(t *testing.T) {