feat(agent): support inline subagent consultation

tests/agent/tools/test_subagent_tools.py

@@ -19,6 +19,37 @@ def _runtime(provider: MagicMock, model: str = "test-model") -> LLMRuntime:
     return LLMRuntime.capture(provider, model, context_window_tokens=128_000)
 
 
+@pytest.mark.asyncio
+async def test_run_inline_returns_result_without_announcement(tmp_path):
+    """Inline subagents return directly instead of injecting a follow-up."""
+    from nanobot.agent.subagent import SubagentManager
+    from nanobot.bus.queue import MessageBus
+
+    provider = MagicMock()
+    manager = SubagentManager(
+        workspace=tmp_path,
+        bus=MessageBus(),
+        max_tool_result_chars=_MAX_TOOL_RESULT_CHARS,
+    )
+    manager.runner.run = AsyncMock(return_value=SimpleNamespace(
+        stop_reason="done",
+        final_content="review result",
+        error=None,
+        tool_events=[],
+    ))
+    manager._announce_result = AsyncMock()
+
+    result = await manager.run_inline(
+        task="review this",
+        session_key="test:c1",
+        runtime=_runtime(provider),
+    )
+
+    assert result == "review result"
+    manager._announce_result.assert_not_awaited()
+    assert manager._task_statuses == {}
+
+
 @pytest.mark.asyncio
 async def test_subagent_exec_tool_receives_allowed_env_keys(tmp_path):
     """allowed_env_keys from ExecToolConfig must be forwarded to the subagent's ExecTool."""
@@ -200,6 +231,40 @@ async def test_spawn_tool_rejects_when_at_concurrency_limit(tmp_path):
     await asyncio.gather(*mgr._running_tasks.values(), return_exceptions=True)
 
 
+@pytest.mark.asyncio
+async def test_spawn_tool_waits_for_inline_result():
+    from nanobot.agent.tools.context import RequestContext, request_context
+    from nanobot.agent.tools.spawn import SpawnTool
+
+    class Manager:
+        max_concurrent_subagents = 1
+
+        def __init__(self):
+            self.inline = AsyncMock(return_value="review result")
+            self.spawn = AsyncMock(return_value="started")
+
+        def get_running_count(self):
+            return 0
+
+        async def run_inline(self, **kwargs):
+            return await self.inline(**kwargs)
+
+    manager = Manager()
+    tool = SpawnTool(manager)
+    runtime = _runtime(MagicMock())
+    with request_context(RequestContext(
+        channel="test",
+        chat_id="c1",
+        session_key="test:c1",
+        runtime=runtime,
+    )):
+        result = await tool.execute(task="review this", wait=True)
+
+    assert result == "review result"
+    manager.inline.assert_awaited_once()
+    manager.spawn.assert_not_awaited()
+
+
 def test_subagent_default_max_concurrent_matches_agent_defaults(tmp_path):
     """Direct SubagentManager construction should use the agent default concurrency limit."""
     from nanobot.agent.subagent import SubagentManager