test: AgentLoop + GovernedLLMClient 集成测试

验证 core/agent/ 通过 LLMProvider Protocol 与 adapters/ 端到端协作。

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

tests/integration/test_agent_governed_e2e.py

@@ -0,0 +1,130 @@
+"""AgentLoop + GovernedLLMClient 集成测试。
+
+验证 core/agent/ 通过 LLMProvider Protocol 与 adapters/llm.py 端到端协作:
+- GovernedLLMClient 满足 LLMProvider Protocol
+- AgentLoop 通过 GovernedLLMClient 完成 search + submit_answer 两步推理
+- token 用量正确累加
+- 遥测数据正确写入 SQLite
+"""
+
+from __future__ import annotations
+
+import json
+from pathlib import Path
+from typing import Any
+from unittest.mock import patch
+
+import pytest
+
+from adapters.breaker import CircuitBreaker
+from adapters.llm import GovernedLLMClient
+from adapters.telemetry import SQLiteTelemetryRecorder
+from core.agent.loop import AgentLoop
+from core.protocols import LLMProvider
+
+
+class _StubDispatcher:
+    """测试用工具调度器，支持 search_tree 和 submit_answer。"""
+
+    async def dispatch(
+        self, tool_name: str, args: dict[str, Any], *, context: dict[str, Any]
+    ) -> str:
+        if tool_name == "submit_answer":
+            return "答案已提交"
+        if tool_name == "search_tree":
+            return "搜索结果: L2-3 节点"
+        raise ValueError(f"未知工具: {tool_name}")
+
+
+def _streaming_result(content: str, thinking: str = "") -> tuple:
+    """构造 _call_streaming 的返回值。
+
+    参数:
+        content: 模型输出的文本内容。
+        thinking: 模型思考过程文本。
+
+    返回:
+        (content, thinking, ttft_ms, max_inter_token_ms, usage_dict) 五元组。
+    """
+    return (
+        content,
+        thinking,
+        50.0,   # ttft_ms
+        10.0,   # max_inter_token_ms
+        {"prompt_tokens": 10, "completion_tokens": 5},
+    )
+
+
+@pytest.mark.asyncio
+async def test_agent_loop_with_governed_client(tmp_path: Path) -> None:
+    """AgentLoop 通过 GovernedLLMClient 完成搜索+提交。"""
+    telemetry = SQLiteTelemetryRecorder(db_path=tmp_path / "telemetry.db")
+    client = GovernedLLMClient(
+        model="test-model",
+        base_url="https://api.test.com/v1",
+        api_key="sk-test",
+        provider="deepseek",
+        thinking=True,
+        breaker=CircuitBreaker(fail_threshold=5, cooldown_s=60.0),
+        cache=None,
+        telemetry=telemetry,
+        timeout_s=30.0,
+        ttft_timeout_s=10.0,
+        inter_token_timeout_s=5.0,
+        max_retries=1,
+        retry_base_delay_s=0.01,
+        retry_max_delay_s=0.05,
+    )
+
+    # 验证 GovernedLLMClient 满足 LLMProvider Protocol
+    assert isinstance(client, LLMProvider)
+
+    call_count = 0
+    responses = [
+        _streaming_result(
+            json.dumps(
+                {
+                    "reflect": {},
+                    "plan": {},
+                    "action": {"tool": "search_tree", "args": {"query": "什么是AI"}},
+                },
+                ensure_ascii=False,
+            ),
+            thinking="让我思考一下",
+        ),
+        _streaming_result(
+            json.dumps(
+                {
+                    "reflect": {},
+                    "plan": {},
+                    "action": {"tool": "submit_answer", "args": {"answer": "AI是..."}},
+                },
+                ensure_ascii=False,
+            ),
+        ),
+    ]
+
+    async def mock_call_streaming(messages: list[dict[str, Any]]) -> tuple:
+        nonlocal call_count
+        result = responses[call_count]
+        call_count += 1
+        return result
+
+    loop = AgentLoop(llm=client, max_steps=10)
+
+    with patch.object(client, "_call_streaming", side_effect=mock_call_streaming):
+        result = await loop.run(
+            system_prompt="你是一个搜索助手",
+            user_prompt="什么是人工智能？",
+            tool_dispatcher=_StubDispatcher(),
+            session_id="test-session",
+        )
+
+    assert result.stop_reason == "finished"
+    assert result.result == {"answer": "AI是..."}
+    assert result.steps_used == 2
+    assert result.steps[0].thought == "让我思考一下"
+    assert result.steps[0].tool_call["tool"] == "search_tree"
+    assert result.steps[1].tool_call["tool"] == "submit_answer"
+    assert result.token_usage["prompt_tokens"] == 20  # 10 * 2
+    assert result.token_usage["completion_tokens"] == 10  # 5 * 2