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/beaver/tools/builtins/skills_admin.py

@@ -0,0 +1,87 @@
+"""Runtime tools for listing and managing skills."""
+
+from __future__ import annotations
+
+from dataclasses import dataclass
+import json
+from typing import Any
+
+from beaver.tools.base import BaseTool, ToolContext, ToolResult, ToolSpec
+
+
+def _result(tool_name: str, success: bool, **payload: Any) -> ToolResult:
+    return ToolResult(
+        success=success,
+        tool_name=tool_name,
+        content=json.dumps({"success": success, **payload}, ensure_ascii=False, indent=2),
+        error=None if success else str(payload.get("error") or "failed"),
+    )
+
+
+@dataclass(slots=True)
+class SkillsListTool(BaseTool):
+    @property
+    def spec(self) -> ToolSpec:
+        return ToolSpec(
+            name="skills_list",
+            description="List available skills with descriptions.",
+            input_schema={"type": "object", "properties": {}},
+            toolset="skills",
+        )
+
+    async def invoke(self, arguments: dict[str, Any], context: ToolContext) -> ToolResult:
+        loader = context.get("skills_loader")
+        if loader is None:
+            return _result(self.spec.name, False, error="skills_loader is unavailable")
+        skills = [
+            {
+                "name": record.name,
+                "description": record.description,
+                "source": record.source,
+                "version": record.version,
+                "tool_hints": list(record.tool_hints),
+            }
+            for record in loader.list_skills(filter_unavailable=False)
+        ]
+        return _result(self.spec.name, True, skills=skills)
+
+
+@dataclass(slots=True)
+class SkillManageTool(BaseTool):
+    @property
+    def spec(self) -> ToolSpec:
+        return ToolSpec(
+            name="skill_manage",
+            description="Create a new skill draft. Publishing still goes through the normal review/publish APIs.",
+            input_schema={
+                "type": "object",
+                "properties": {
+                    "action": {"type": "string", "enum": ["create_draft"]},
+                    "name": {"type": "string"},
+                    "description": {"type": "string"},
+                    "content": {"type": "string"},
+                },
+                "required": ["action", "name", "content"],
+            },
+            toolset="skills",
+        )
+
+    async def invoke(self, arguments: dict[str, Any], context: ToolContext) -> ToolResult:
+        if arguments.get("action") != "create_draft":
+            return _result(self.spec.name, False, error="only create_draft is supported")
+        draft_service = context.get("draft_service")
+        if draft_service is None:
+            return _result(self.spec.name, False, error="draft_service is unavailable")
+        name = str(arguments.get("name") or "").strip()
+        content = str(arguments.get("content") or "").strip()
+        if not name or not content:
+            return _result(self.spec.name, False, error="name and content are required")
+        draft = draft_service.create_new_skill_draft(
+            skill_name=name,
+            proposed_content=content,
+            proposed_frontmatter={"description": str(arguments.get("description") or name)},
+            created_by=context.user_id or "agent",
+            reason="created by skill_manage tool",
+            trigger_session_id=context.session_id,
+        )
+        return _result(self.spec.name, True, draft=draft.to_dict())