feat(coordinator): 添加团队节点默认最大工具迭代次数配置

添加 DEFAULT_TEAM_NODE_MAX_TOOL_ITERATIONS 配置项以控制团队节点的最大工具迭代次数，
并修改 LocalAgentRunner 中的逻辑来使用此默认值当 envelope 中未指定时。

fix(runtime): 修复团队节点运行成功判断逻辑

更新运行成功判断条件，将 finish_reason 为 "max_tool_iterations_finalized" 的情况
视为运行失败，并添加对原始工具调用输出的检测，避免将其误判为成功完成。

feat(mcp): 添加团队工作流MCP工具类别支持

增加新的本地MCP工具类别 "team_workflow" 及其对应的工具创建功能，
为团队工作流提供本地工具支持。

refactor(engine): 调整AgentLoop最大工具迭代次数设置

将 AgentProfile 中的默认 max_tool_iterations 从 30 增加到 100，
同时移除 TaskExecutionPlanner 构造函数中的重复参数传递。

perf(mcp): 优化MCP连接管理避免重复连接

添加 mcp_connected 标志来跟踪MCP连接状态，确保 connect_all 只执行一次，
提高性能并避免不必要的重复连接。

refactor(skills): 移除技能团队模板相关功能

移除与技能团队模板相关的代码，包括解析、存储和处理逻辑，
简化技能记录结构和加载流程。

feat(process): 增强会话过程投影器功能

添加技能激活快照事件处理，改进团队运行完成消息显示，
并增强技能激活事件的时间戳记录功能。

refactor(tasks): 简化任务尝试编排器团队执行逻辑

移除团队执行相关代码，将所有任务统一按单步执行处理，
简化任务编排器的复杂度并提升执行效率。

fix(evidence): 修复节点证据评估中需求验证逻辑

更新节点证据评估逻辑，跳过自然语言证据需求的确定性验证，
只执行机器可读的需求验证，避免因自然语言需求导致的节点失败。

app-instance/backend/beaver/coordinator/local.py

@@ -9,6 +9,7 @@ from beaver.engine.providers import ProviderBundle
 from beaver.tasks.evidence import EvidenceBuilder, evaluate_node_evidence
 
 from .models import DelegationEnvelope, NodeRunResult
+from .runtime_defaults import DEFAULT_TEAM_NODE_MAX_TOOL_ITERATIONS
 
 
 class LocalAgentRunner:
@@ -55,7 +56,11 @@ class LocalAgentRunner:
             pinned_skill_names=envelope.inherited_pinned_skills,
             pinned_skill_contexts=envelope.inherited_pinned_skill_contexts,
             allowed_tool_names=envelope.allowed_tool_names,
-            max_tool_iterations=envelope.max_tool_iterations,
+            max_tool_iterations=(
+                envelope.max_tool_iterations
+                if envelope.max_tool_iterations is not None
+                else DEFAULT_TEAM_NODE_MAX_TOOL_ITERATIONS
+            ),
             allow_candidate_generation=allow_candidate_generation,
         )
         loaded = target_loop.boot()
@@ -70,7 +75,8 @@ class LocalAgentRunner:
             envelope.required_evidence,
             result.output_text,
         )
-        run_succeeded = result.finish_reason == "stop"
+        raw_tool_call_output = self._looks_like_raw_tool_call(result.output_text)
+        run_succeeded = result.finish_reason in {"stop", "max_tool_iterations_finalized"} and not raw_tool_call_output
         if not run_succeeded:
             completion_status = "failed"
         elif evidence_gaps:
@@ -81,7 +87,10 @@ class LocalAgentRunner:
         if completion_status == "partial":
             error = "; ".join(evidence_gaps)
         else:
-            error = None if success else (result.output_text or result.finish_reason)
+            if raw_tool_call_output:
+                error = "finalized output is a raw tool call"
+            else:
+                error = None if success else (result.output_text or result.finish_reason)
         return NodeRunResult(
             node_id=envelope.node_id or envelope.agent.name,
             success=success,
@@ -169,3 +178,16 @@ class LocalAgentRunner:
             "If no published skill matches, return [] and let the node continue without skills."
         )
         return "\n\n".join(sections)
+
+    @staticmethod
+    def _looks_like_raw_tool_call(output_text: str | None) -> bool:
+        text = (output_text or "").strip()
+        if not text:
+            return False
+        markers = (
+            "<｜｜DSML｜｜tool_calls>",
+            "<｜｜DSML｜｜invoke",
+            "<tool_call",
+            "<function=",
+        )
+        return any(marker in text for marker in markers)

app-instance/backend/beaver/coordinator/runtime_defaults.py

@@ -0,0 +1,3 @@
+"""Runtime defaults shared by Beaver team planning and execution."""
+
+DEFAULT_TEAM_NODE_MAX_TOOL_ITERATIONS = 100