diff --git a/evals/token_accounting/display_provenance.py b/evals/token_accounting/display_provenance.py
index 391872b578..527412a929 100644
--- a/evals/token_accounting/display_provenance.py
+++ b/evals/token_accounting/display_provenance.py
@@ -77,8 +77,10 @@ def child(out: Path) -> None:
             cli.conversation_history = result["messages"]
         else:
             cli.conversation_history.append({"role": "user", "content": "new unpriced question"})
+        agent._session_messages = cli.conversation_history
         print(f"\n=== {scenario}: CLI /context ===", flush=True)
         cli.process_command("/context")
+        print("CLI status bar:", cli._build_status_bar_text(width=120), flush=True)
         usage = _get_usage(agent)
         session = {"agent": agent, "history": cli.conversation_history, "history_lock": threading.RLock()}
         print("=== TUI/Desktop live /context ===", flush=True)
@@ -89,6 +91,8 @@ def child(out: Path) -> None:
         print(asyncio.run(runner._handle_status_command(MessageEvent(text="/status", source=source, message_id="fixture"))), flush=True)
         print("=== gateway /context ===", flush=True)
         print(asyncio.run(runner._handle_context_command(MessageEvent(text="/context", source=source, message_id="fixture"))), flush=True)
+        print("=== gateway /usage category block ===", flush=True)
+        print("\n".join(runner._context_breakdown_lines(agent, source)), flush=True)
         results[scenario] = {"breakdown": compute_session_context_breakdown(agent, cli.conversation_history), "usage": usage}
     (out / "payloads.json").write_text(json.dumps(results, indent=2))
     server.omit_usage = True
diff --git a/gateway/slash_commands_status.py b/gateway/slash_commands_status.py
index e4f85a3995..e449018c23 100644
--- a/gateway/slash_commands_status.py
+++ b/gateway/slash_commands_status.py
@@ -118,6 +118,8 @@ def _context_compressor_lines(agent, ctx, used: int) -> list[str]:
     """/context full view: auto-compression threshold/headroom, compression count + last savings,
     and cumulative throughput (labelled as throughput, NOT context size)."""
     lines: list[str] = []
+    from agent.context_breakdown import context_display_source
+    mark = "~" if context_display_source(ctx) != "provider_usage" else ""
     threshold = _n(ctx, "threshold_tokens")
     threshold_pct = f"{_n(ctx, 'threshold_percent') * 100:.0f}"
     if threshold > 0:
@@ -126,7 +128,7 @@ def _context_compressor_lines(agent, ctx, used: int) -> list[str]:
                            threshold_pct=threshold_pct))
         else:
             lines.append(t("gateway.context.threshold", threshold=_fmt(threshold),
-                           threshold_pct=threshold_pct, to_go=_fmt(threshold - used)))
+                           threshold_pct=threshold_pct, to_go=mark + _fmt(threshold - used)))
     compressions = _n(ctx, "compression_count")
     lines.append(t("gateway.context.compressions", count=compressions))
     savings = getattr(ctx, "_last_compression_savings_pct", None) if compressions else None
@@ -501,7 +503,7 @@ class GatewayStatusCommandsMixin:
                 if label.endswith(f"breakdown_cat_{cat_id}"):  # missing key: t() echoes it back
                     label = str(cat.get("label") or cat_id)
                 pct = round(tokens / total * 100) if total else 0
-                out.append(t("gateway.usage.breakdown_line", label=label, count="~" + _fmt(tokens), pct=f"~{pct}"))
+                out.append(t("gateway.usage.breakdown_line", label=label, count=_fmt(tokens), pct=f"~{pct}"))
             return out if len(out) > 1 else []
         except Exception:
             return []