feat: 工具系统迁移 + 重型插件骨架 + 前端交互增强

- 工具系统从 kilostar/plugin/tool_plugin/ 迁移到 data/toolset/（manifest.json 声明式）
- 新增 plugin_runtime 模块：BaseOrganization / GlobalPluginManager / loader / tool_bridge
- 新增 org_task + org_task_event 表及 DAO（alembic 0009）
- 新增 /api/v1/plugin 路由（submit/status/stream/install/reload）
- 新增 data/plugin/example_dept 示例重型插件
- regulatory_node 支持聊天历史上下文注入
- send_file 改为 artifact 存盘 + SSE 推送下载链接
- 前端 WorkflowFileCard 组件 + ToolSettings README 渲染
- utils 整理：合并 access/role_check、standalone_proxy→ray_compat、删除废弃模块
- 项目结构文档移至 docs/STRUCTURE.md 并详细展开

Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>

kilostar/core/individual/consciousness_node/consciousness_node.py

@@ -14,7 +14,7 @@
 
 
 from typing import Union, overload
-from kilostar.utils.standalone_proxy import actor_class
+from kilostar.utils.ray_compat import actor_class
 from kilostar.core.individual.consciousness_node.template import (
     ConsciousnessNodeDeps,
     ForregulatoryNode,
@@ -29,7 +29,7 @@ from kilostar.core.global_state_machine.global_state_machine import GlobalStateM
 from kilostar.core.global_state_machine.model_provider.base_provider import Provider
 from kilostar.adapter.model_adapter.agent_factory import AgentFactory
 from kilostar.utils.ray_hook import ray_actor_hook
-from kilostar.utils.i18n import agent_prompt
+from kilostar.utils.prompts import agent_prompt
 
 
 @actor_class

kilostar/core/individual/control_node/control_node.py

@@ -13,7 +13,7 @@
 #  limitations under the License.
 
 from pydantic_ai import Agent, RunContext
-from kilostar.utils.standalone_proxy import actor_class
+from kilostar.utils.ray_compat import actor_class
 from kilostar.core.global_state_machine.global_state_machine import GlobalStateMachine
 from kilostar.core.global_state_machine.model_provider.base_provider import Provider
 from kilostar.adapter.model_adapter.agent_factory import AgentFactory
@@ -22,7 +22,7 @@ from kilostar.core.individual.control_node.template import (
     ForWorkflowInput,
     ControlNodeDeps,
 )
-from kilostar.utils.i18n import agent_prompt
+from kilostar.utils.prompts import agent_prompt
 
 
 @actor_class

kilostar/core/individual/regulatory_node/regulatory_node.py

@@ -15,7 +15,7 @@
 import asyncio
 import datetime
 from typing import Union
-from kilostar.utils.standalone_proxy import actor_class
+from kilostar.utils.ray_compat import actor_class
 from kilostar.adapter.model_adapter.agent_factory import AgentFactory
 from kilostar.core.global_state_machine.global_state_machine import GlobalStateMachine
 from kilostar.core.global_state_machine.model_provider import Provider
@@ -25,7 +25,7 @@ from kilostar.core.individual.regulatory_node.template import (
     MessageResponse
 )
 from pydantic_ai import RunContext, Agent
-from kilostar.utils.i18n import agent_prompt
+from kilostar.utils.prompts import agent_prompt
 
 
 @actor_class
@@ -111,15 +111,20 @@ class RegulatoryNode:
                 )
             return prompt
 
-    async def working(self, payload: MessageRequest) -> Union[MessageResponse, None]:
+    async def working(
+        self,
+        payload: MessageRequest,
+        message_history: list | None = None,
+    ) -> Union[MessageResponse, None]:
         """working方法，是节点唯一的调用方法，对_run函数的结果进行判断并返回最终回复
         Args:
             payload: 消息载荷，包含所有信息
+            message_history: pydantic-ai ``ModelMessage`` 列表，传入历史让多轮对话连贯
 
         Returns:
             MessageResponse 或 None，监控节点对用户的结构化回复
         """
-        return await self._run(payload)
+        return await self._run(payload, message_history=message_history)
 
     _CHAT_INSTRUCTIONS = (
         "你是 kilostar 智能助手。你现在处于【直接对话模式】，请直接回答用户的问题。\n"
@@ -130,7 +135,12 @@ class RegulatoryNode:
         "4. 回复应当完整、有帮助，避免过于简短。\n"
     )
 
-    async def stream_working(self, payload: MessageRequest, token_queue: "asyncio.Queue") -> None:
+    async def stream_working(
+        self,
+        payload: MessageRequest,
+        token_queue: "asyncio.Queue",
+        message_history: list | None = None,
+    ) -> None:
         """流式对话：完整执行 agent graph（含工具调用），逐 token 推送文本到 queue。
 
         使用 event_stream_handler 回调拿到每个 text delta，保证工具调用后
@@ -167,6 +177,7 @@ class RegulatoryNode:
                 output_type=str,
                 instructions=self._CHAT_INSTRUCTIONS,
                 event_stream_handler=_stream_handler,
+                message_history=message_history,
             )
         except Exception as e:
             self.logger.exception(f"RegulatoryNode.stream_working failed: {e}")
@@ -175,7 +186,9 @@ class RegulatoryNode:
             await token_queue.put(None)
 
     async def _run(
-        self, payload: MessageRequest
+        self,
+        payload: MessageRequest,
+        message_history: list | None = None,
     ) -> Union[MessageResponse, None]:
         platform = payload.platform
         user_name = payload.user_name
@@ -187,8 +200,11 @@ class RegulatoryNode:
                 user_name=user_name,
                 time=time_str
             )
-            agent_response = await self.agent.run(user_prompt=message,
-                                            deps=deps,)
+            agent_response = await self.agent.run(
+                user_prompt=message,
+                deps=deps,
+                message_history=message_history,
+            )
             response: MessageResponse = agent_response.output
             response.platform = platform
             response.platform_id = payload.platform_id