feat(desktop): polish desktop shell and shared WebUI surfaces (#4195)

* feat(desktop): add native host scaffold

* feat(webui): track turns and usage in gateway

* feat(webui): polish desktop chat experience

* feat(apps): add ArcGIS and Joplin logos

* feat(desktop): polish shell and shared surfaces

* fix(webui): avoid preview chips for glob references

* test: align CI expectations for token fallback

* feat(webui): preview prompt rail entries

* feat(webui): add prompt navigator drawer

* style(webui): refine prompt navigator placement

* style(webui): align prompt navigator with header actions

* style(webui): simplify prompt navigator header

* refactor(webui): clean thread resource refresh

* feat(desktop): add native reply notifications

* fix(webui): preserve desktop restart and replay state

* fix(desktop): harden gateway proxy startup

* fix(web): fall back when readability is unavailable

* fix(desktop): hide window instead of closing on macos

* fix(webui): unify desktop header actions

* fix(webui): simplify prompt history rows

* fix(desktop): log notification delivery failures

* chore(desktop): clean source package artifacts

* fix(cron): support one-time relative reminders

* fix(webui): reveal scroll button in place

* Revert "fix(cron): support one-time relative reminders"

This reverts commit 4c4661da120a3c7283e0768412bae48604e7390b.

* refactor(webui): extract token usage heatmap

* docs(desktop): clarify contributor guides

---------

Co-authored-by: chengyongru <2755839590@qq.com>

tests/agent/test_dream.py

@@ -356,7 +356,6 @@ class TestEphemeralHooks:
         await loop.process_direct("test", session_key="cli:normal")
         spy.before_iteration.assert_called()
 
-
 class TestDreamCommitMessage:
     async def test_commit_includes_response_summary(self, tmp_path):
         """Git auto-commit after Dream should include the LLM response in the body."""

tests/agent/test_loop_save_turn.py

@@ -592,16 +592,16 @@ async def test_internal_continuation_queues_turn_without_fake_user_history(
                 "paused",
                 [],
                 [*initial_messages, {"role": "assistant", "content": "paused"}],
-                "max_iterations",
-                False,
-            )
+                    "max_iterations",
+                    False,
+                )
         return (
             "done",
             [],
             [*initial_messages, {"role": "assistant", "content": "done"}],
-            "completed",
-            False,
-        )
+                "completed",
+                False,
+            )
 
     loop._run_agent_loop = fake_run_agent_loop  # type: ignore[method-assign]
     pending: asyncio.Queue[InboundMessage] = asyncio.Queue()
@@ -665,9 +665,9 @@ async def test_internal_continuation_preserves_streaming_route_metadata(
                 "paused",
                 [],
                 [*initial_messages, {"role": "assistant", "content": "paused"}],
-                "max_iterations",
-                False,
-            )
+                    "max_iterations",
+                    False,
+                )
         assert on_stream is not None
         assert on_stream_end is not None
         await on_stream("done")
@@ -744,9 +744,9 @@ async def test_websocket_internal_continuation_keeps_single_visible_run(
                 "paused",
                 [],
                 [*initial_messages, {"role": "assistant", "content": "paused"}],
-                "max_iterations",
-                False,
-            )
+                    "max_iterations",
+                    False,
+                )
         return (
             "done",
             [],

tests/agent/test_runner_hooks.py

@@ -170,6 +170,48 @@ async def test_runner_passes_cached_tokens_to_hook_context():
 
     assert len(captured_usage) == 1
     assert captured_usage[0]["cached_tokens"] == 150
+    assert captured_usage[0]["provider_tokens"] == 220
+
+
+@pytest.mark.asyncio
+async def test_runner_estimates_usage_when_provider_omits_usage(monkeypatch):
+    from nanobot.agent.hook import AgentHook, AgentHookContext
+    from nanobot.agent.runner import AgentRunner, AgentRunSpec
+
+    provider = MagicMock(spec=LLMProvider)
+    captured_usage: list[dict] = []
+
+    class UsageHook(AgentHook):
+        async def after_iteration(self, context: AgentHookContext) -> None:
+            captured_usage.append(dict(context.usage))
+
+    async def chat_with_retry(**kwargs):
+        return LLMResponse(content="done", tool_calls=[], usage={})
+
+    provider.chat_with_retry = chat_with_retry
+    tools = MagicMock()
+    tools.get_definitions.return_value = [{"type": "function", "function": {"name": "lookup"}}]
+    monkeypatch.setattr(
+        "nanobot.agent.runner.estimate_prompt_tokens_chain",
+        lambda provider, model, messages, tools: (123, "test"),
+    )
+    monkeypatch.setattr("nanobot.agent.runner.estimate_message_tokens", lambda message: 7)
+
+    runner = AgentRunner(provider)
+    result = await runner.run(AgentRunSpec(
+        initial_messages=[{"role": "user", "content": "hi"}],
+        tools=tools,
+        model="test-model",
+        max_iterations=1,
+        max_tool_result_chars=_MAX_TOOL_RESULT_CHARS,
+        hook=UsageHook(),
+    ))
+
+    assert result.usage["prompt_tokens"] == 123
+    assert result.usage["completion_tokens"] == 7
+    assert result.usage["total_tokens"] == 130
+    assert result.usage["estimated_tokens"] == 130
+    assert captured_usage[0]["estimated_tokens"] == 130
 
 
 @pytest.mark.asyncio
@@ -232,7 +274,12 @@ async def test_runner_calls_run_level_hooks_on_success():
             "done",
             "completed",
             None,
-            {"prompt_tokens": 3, "completion_tokens": 2},
+            {
+                "prompt_tokens": 3,
+                "completion_tokens": 2,
+                "total_tokens": 5,
+                "provider_tokens": 5,
+            },
             ["user", "assistant"],
         ),
         ("on_finally", "completed", None),