feat(engine): 添加MCP连接管理和工具集成功能

- 集成MCP连接管理器，支持MCP服务器连接
- 添加多种内置工具：ClarifyTool、CronTool、DelegateTool、ExecuteCodeTool、
  PatchFileTool、ProcessTool、SendMessageTool、SpawnTool、TerminalTool、
  TodoTool、WebFetchTool、WebSearchTool、WriteFileTool等
- 实现工具注册和装配功能
- 添加技能选择上下文参数
- 支持思考模式控制参数thinking_enabled

feat(coordinator): 重构任务执行计划器参数命名

- 将learning_candidate_enabled重命名为allow_candidate_generation
- 更新TeamGraphScheduler中的参数传递
- 修改LocalAgentRunner中的相关参数处理
- 更新README文档中的相应描述

refactor(context): 标准化工具调用参数格式

- 添加_json导入用于参数序列化
- 实现_provider_tool_calls方法标准化OpenAI兼容的工具调用载荷
- 修复工具调用中参数非字符串类型的序列化问题

refactor(session): 优化消息历史记录过滤逻辑

- 修改get_messages_as_conversation为基于运行状态过滤消息
- 排除未完成、失败或错误结束的运行记录
- 改进对话历史的可见性控制机制

fix(store): 修复FTS索引重建逻辑

- 添加异常处理防止FTS索引创建失败
- 实现_rebuild_fts_index方法重新构建全文搜索索引
- 优化索引触发器和表的维护流程

app-instance/backend/tests/unit/test_skill_assembler.py

@@ -0,0 +1,157 @@
+from __future__ import annotations
+
+import asyncio
+from types import SimpleNamespace
+
+from beaver.engine.providers.base import LLMProvider, LLMResponse
+from beaver.skills.assembler.task_assembler import SkillAssembler
+
+
+class RecordingProvider(LLMProvider):
+    def __init__(self) -> None:
+        super().__init__()
+        self.thinking_enabled: bool | None = None
+
+    async def chat(
+        self,
+        messages: list[dict],
+        tools: list[dict] | None = None,
+        model: str | None = None,
+        max_tokens: int = 4096,
+        temperature: float = 0.7,
+        thinking_enabled: bool | None = None,
+    ) -> LLMResponse:
+        self.thinking_enabled = thinking_enabled
+        return LLMResponse(content='["daily-news"]', provider_name="stub", model="stub-model")
+
+    def get_default_model(self) -> str:
+        return "stub-model"
+
+
+class SequencedProvider(LLMProvider):
+    def __init__(self, responses: list[str]) -> None:
+        super().__init__()
+        self.responses = list(responses)
+        self.messages: list[list[dict]] = []
+
+    async def chat(
+        self,
+        messages: list[dict],
+        tools: list[dict] | None = None,
+        model: str | None = None,
+        max_tokens: int = 4096,
+        temperature: float = 0.7,
+        thinking_enabled: bool | None = None,
+    ) -> LLMResponse:
+        self.messages.append(messages)
+        content = self.responses.pop(0)
+        return LLMResponse(content=content, provider_name="stub", model="stub-model")
+
+    def get_default_model(self) -> str:
+        return "stub-model"
+
+
+class StaticRetriever:
+    async def retrieve(self, **kwargs):
+        return kwargs["candidates"][: kwargs["top_k"]]
+
+
+class LoaderWithFullSkill:
+    def build_selection_candidates(self) -> list[dict[str, str]]:
+        return [
+            {
+                "name": "docker-debug",
+                "description": "General container tips.",
+                "version": "v1",
+                "content_hash": "abc",
+            }
+        ]
+
+    def load_published_skill(self, name: str) -> str | None:
+        if name != "docker-debug":
+            return None
+        return """---
+description: General container tips.
+tools:
+  - search_files
+---
+
+# Docker Debug
+
+Use this skill when doing Docker log triage and container failure analysis.
+"""
+
+    def get_skill_record(self, name: str):
+        return SimpleNamespace(version="v1", content_hash="abc", tool_hints=["search_files"])
+
+
+def test_skill_selection_receives_thinking_mode() -> None:
+    provider = RecordingProvider()
+    assembler = SkillAssembler(loader=SimpleNamespace())
+
+    selected = asyncio.run(
+        assembler._select_skill_names(
+            task_description="summarize daily news",
+            candidates=[{"name": "daily-news", "description": "Summarize news"}],
+            provider=provider,
+            model="Qwen3.6-35B",
+            thinking_enabled=False,
+        )
+    )
+
+    assert selected == ["daily-news"]
+    assert provider.thinking_enabled is False
+
+
+def test_skill_assembler_loads_detail_directly_for_small_candidate_sets() -> None:
+    provider = SequencedProvider(['["docker-debug"]'])
+    assembler = SkillAssembler(loader=LoaderWithFullSkill(), retriever=StaticRetriever())
+
+    result = asyncio.run(
+        assembler.assemble(
+            task_description="debug a failing Docker container",
+            provider=provider,
+            model="stub-model",
+        )
+    )
+
+    assert [skill.name for skill in result.activated_skills] == ["docker-debug"]
+    assert result.activated_skills[0].tool_hints == ["search_files"]
+    assert [item["stage"] for item in result.llm_interactions] == ["final"]
+    assert len(provider.messages) == 1
+    first_user_prompt = provider.messages[0][1]["content"]
+    assert "Use this skill when doing Docker log triage" in first_user_prompt
+
+
+def test_skill_assembler_shortlists_before_loading_detail_for_large_candidate_sets() -> None:
+    provider = SequencedProvider(['["docker-debug"]', '["docker-debug"]'])
+    loader = LoaderWithFullSkill()
+    original_candidates = loader.build_selection_candidates
+    loader.build_selection_candidates = lambda: [
+        *original_candidates(),
+        {
+            "name": "other-skill",
+            "description": "Other workflow.",
+            "version": "v1",
+            "content_hash": "def",
+        },
+    ]
+    assembler = SkillAssembler(
+        loader=loader,
+        retriever=StaticRetriever(),
+        max_detailed_candidates=1,
+    )
+
+    result = asyncio.run(
+        assembler.assemble(
+            task_description="debug a failing Docker container",
+            provider=provider,
+            model="stub-model",
+        )
+    )
+
+    assert [skill.name for skill in result.activated_skills] == ["docker-debug"]
+    assert [item["stage"] for item in result.llm_interactions] == ["shortlist", "final"]
+    assert len(provider.messages) == 2
+    assert "Use this skill when doing Docker log triage" not in provider.messages[0][1]["content"]
+    assert "Use this skill when doing Docker log triage" in provider.messages[1][1]["content"]