fix(interface): clarify cached vs new input token breakdown

strix/interface/utils.py

@@ -296,16 +296,26 @@ def _build_llm_usage_stats(
     input_tokens = _int_stat(usage, "input_tokens")
     output_tokens = _int_stat(usage, "output_tokens")
     cached_tokens = _detail_value(usage, "input_tokens_details", "cached_tokens")
+    uncached_tokens = max(0, input_tokens - cached_tokens)
     cost = _float_stat(usage, "cost")
 
     stats_text.append("\n")
-    stats_text.append("Input Tokens ", style="dim")
-    stats_text.append(format_token_count(input_tokens), style="white")
-
-    if live or cached_tokens > 0:
+    if cached_tokens > 0:
+        stats_text.append("Input (total) ", style="dim")
+        stats_text.append(format_token_count(input_tokens), style="white")
         stats_text.append("  ·  ", style="dim white")
-        stats_text.append("Cached Tokens ", style="dim")
+        stats_text.append("Cached (in total) ", style="dim")
         stats_text.append(format_token_count(cached_tokens), style="white")
+        stats_text.append("  ·  ", style="dim white")
+        stats_text.append("New input ", style="dim")
+        stats_text.append(format_token_count(uncached_tokens), style="white")
+    else:
+        stats_text.append("Input Tokens ", style="dim")
+        stats_text.append(format_token_count(input_tokens), style="white")
+        if live:
+            stats_text.append("  ·  ", style="dim white")
+            stats_text.append("Cached (in total) ", style="dim")
+            stats_text.append(format_token_count(cached_tokens), style="white")
 
     separator = "\n" if live else "  ·  "
     stats_text.append(separator, style="dim white")

tests/test_interface_stats.py

@@ -0,0 +1,70 @@
+"""Tests for TUI / CLI usage stats rendering."""
+
+from __future__ import annotations
+
+from types import SimpleNamespace
+from unittest.mock import MagicMock
+
+from strix.interface.utils import build_final_stats_text, build_live_stats_text
+
+
+def _report_state_with_usage(usage: dict) -> SimpleNamespace:
+    state = SimpleNamespace(vulnerability_reports=[])
+    state.get_total_llm_usage = MagicMock(return_value=usage)
+    return state
+
+
+def test_usage_stats_without_cache_uses_simple_input_label() -> None:
+    text = build_final_stats_text(
+        _report_state_with_usage(
+            {
+                "requests": 1,
+                "input_tokens": 1_000_000,
+                "output_tokens": 500_000,
+                "cost": 0.5,
+            }
+        )
+    )
+
+    plain = text.plain
+    assert "Input Tokens 1.0M" in plain
+    assert "Cached" not in plain
+    assert "New input" not in plain
+
+
+def test_live_mode_without_cache_still_shows_cached_row() -> None:
+    """Live sidebar always exposes the cache metric, even when it is still zero."""
+    text = build_live_stats_text(
+        _report_state_with_usage(
+            {
+                "requests": 1,
+                "input_tokens": 1_000_000,
+                "output_tokens": 500_000,
+            }
+        )
+    )
+
+    plain = text.plain
+    assert "Input Tokens 1.0M" in plain
+    assert "Cached (in total) 0" in plain
+    assert "New input" not in plain
+    assert plain.index("Input Tokens") < plain.index("Cached (in total)")
+
+
+def test_usage_stats_with_cache_shows_subset_breakdown() -> None:
+    text = build_live_stats_text(
+        _report_state_with_usage(
+            {
+                "requests": 1,
+                "input_tokens": 20_000_000,
+                "output_tokens": 1_000_000,
+                "input_tokens_details": {"cached_tokens": 19_000_000},
+            }
+        )
+    )
+
+    plain = text.plain
+    assert "Input (total) 20.0M" in plain
+    assert "Cached (in total) 19.0M" in plain
+    assert "New input 1.0M" in plain
+    assert plain.index("Input (total)") < plain.index("Cached (in total)") < plain.index("New input")