feat(toolset): 工具系统重构为 toolset 统一管理,新增系统预置工具集

将工具管理从"agent 挂单个 tool"改为"agent 挂 toolset"模式:
- 三个系统预置工具集(system_basic/system_chat/system_workflow)入 DB
- 新增 send_file 工具(系统对话工具集)、修复 approval actor 调用 bug
- 后端 agent 加载全部走 toolset 链路,移除 load_tools_from_list
- 前端工具集中心卡片展示 + agent 配置改为 toolset 多选
- resource API 增加 category 过滤与系统 toolset 保护

Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
This commit is contained in:
2026-06-05 18:03:49 +00:00
parent d39c80743d
commit 0e57c5cf16
23 changed files with 584 additions and 169 deletions
@@ -0,0 +1,60 @@
# Copyright 2026 zhaoxi826
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
import json
from kilostar.plugin.tool_plugin.base_tool import BaseToolData
from kilostar.utils.ray_hook import ray_actor_hook
from typing import List, Literal, Dict
class SendFileToolData(BaseToolData):
"""``send_file`` 工具元数据:把 agent 生成的文件作为附件推送到对话窗口。"""
is_system: bool = True
action_scope: List[
Literal[
"control_node",
"consciousness_node",
"regulatory_node",
"growth_node",
"",
]
] = []
config_args: Dict[str, str] = {}
category: str = "system"
async def send_file(filename: str, content: str, trace_id: str) -> str:
"""把 agent 生成的文件作为附件发送给当前对话窗口。
通过 global_workflow_manager 的 pending 队列推送一条带特殊前缀的 JSON 消息,
前端识别后渲染为可下载的文件卡片。
Args:
filename: 文件名(含扩展名),如 "report.md" / "main.py"
content: 文件内容(UTF-8 文本)
trace_id: 当前会话/工作流的 trace_id
Returns:
发送结果说明
"""
payload = json.dumps(
{"type": "file", "filename": filename, "content": content},
ensure_ascii=False,
)
actor_list = ray_actor_hook("global_workflow_manager")
await actor_list.global_workflow_manager.put_pending.remote(
trace_id, f"__FILE__{payload}"
)
return f"已发送文件: {filename}"