valkryhx/google_adk_agent / skills/agent_team_to_be_update
Agent 团队协作(去中心化拉模型)
Flagged赋予 Agent 作为集群指挥官(Leader)的能力,通过去中心化任务队列将子任务派发给 Worker 节点并发执行。
| File | Kind | Size |
|---|---|---|
| SKILL.md | skill md | 17.9 KB |
| decentralized_tools.py | script | 36.2 KB |
| dependency_analyzer.py | script | 10.9 KB |
| inbox_watcher.py | script | 8.5 KB |
| loop_executor.py | script | 16.4 KB |
| mailbox.py | script | 15.3 KB |
| models.py | script | 9.1 KB |
| path_guard.py | script | 13.3 KB |
| planner.py | script | 14.6 KB |
| polling_daemon.py | script | 7.7 KB |
| self_claim_loop.py | script | 20.1 KB |
| task_queue.py | script | 24.1 KB |
| team_config.py | script | 7.5 KB |
| test/__init__.py | script | 40 B |
| test/test_blog_system_develop.py | script | 12.0 KB |
| test/test_dag_create_validation.py | script | 2.7 KB |
| test/test_decentralized_arch.py | script | 36.9 KB |
| test/test_deepresearch.py | script | 14.6 KB |
| test/test_loop_dag.py | script | 17.0 KB |
| test/test_real_training.py | script | 9.2 KB |
| test/test_tdd_gate.py | script | 4.5 KB |
| tools.py | script | 85.8 KB |
| verification_hooks.py | script | 10.7 KB |
| worktree_manager.py | script | 12.8 KB |
test/test_blog_system_develop.py
View on GitHub#!/usr/bin/env python3
# -*- coding: utf-8 -*-
r"""
================================================================================
Agent Team DAG 测试脚本 - 博客系统开发示例
================================================================================
【使用方式】
## 1. 独立运行测试
直接在命令行运行:
```bash
cd D:\git_codes\google_adk_helloworld_git\skills\agent_team_to_be_update\test
python test_blog_system_develop.py
```
## 2. 在 Swarm 集群中使用
启动集群后 (start_demo_swarm.bat),在 Web 界面输入:
```
测试任务依赖系统:
- 让 Worker 8001 创建用户模块(Schema + API + 前端)
- 让 Worker 8002 创建文章模块(依赖用户模块)
- 让 Worker 8003 创建评论模块(依赖用户和文章模块)
- 最后做集成测试
```
## 3. 导入模块到你的代码
```python
import sys
sys.path.insert(0, 'skills/agent_team_to_be_update')
from models import Task
from task_queue import TaskQueue
from mailbox import Mailbox
from path_guard import PathGuard
from verification_hooks import TeammateIdleHook, TaskCompletedHook
from planner import TaskPlanner
```
================================================================================
【测试内容】
本脚本测试 Claude Code 风格的 Agent Team DAG 任务依赖系统:
1. Task 数据模型 - 支持 blockedBy 依赖声明
2. TaskQueue 任务队列 - 文件锁抢任务机制
3. Wave 执行顺序 - 自动计算可并行执行的任务
4. Mailbox 通信 - Agent 间点对点消息
5. PathGuard 路径守卫 - 防止访问禁止目录
================================================================================
"""
import sys
import os
import tempfile
import shutil
if sys.platform == "win32":
import codecs
if hasattr(sys.stdout, "buffer"):
sys.stdout = codecs.getwriter("utf-8")(sys.stdout.buffer, "strict")
if hasattr(sys.stderr, "buffer"):
sys.stderr = codecs.getwriter("utf-8")(sys.stderr.buffer, "strict")
# 添加模块路径
sys.path.insert(0, os.path.dirname(os.path.dirname(os.path.abspath(__file__))))
from models import Task
from task_queue import TaskQueue
def print_task(task: Task, indent: str = " "):
"""打印任务信息"""
print(f"{indent}Task: {task.name}")
print(f"{indent} ID: {task.id}")
print(f"{indent} Status: {task.status}")
print(f"{indent} Owner: {task.owner or '(unclaimed)'}")
if task.blocked_by:
print(f"{indent} blockedBy: {task.blocked_by}")
print()
def test_basic_dag():
"""测试基础 DAG 功能"""
print("=" * 60)
print("测试 1: 基础 DAG - 博客系统开发")
print("=" * 60)
# 创建临时目录
temp_dir = tempfile.mkdtemp(prefix="agent_team_test_")
print(f"测试目录: {temp_dir}\n")
try:
# 初始化任务队列
queue = TaskQueue(team_id="test_blog", base_dir=temp_dir)
# 创建 DAG 任务结构:
#
# Task 1: 设计 Schema (无依赖)
# |
# Task 2: 实现后端 API (blockedBy: 1)
# |
# Task 3: 实现前端页面 (blockedBy: 2)
# Task 4: 编写测试 (blockedBy: 2)
# |
# Task 5: 集成测试 (blockedBy: 3, 4)
print("创建任务...")
# Task 1: 无依赖
task1 = queue.create_task(
name="设计数据库 Schema",
description="设计用户表、文章表、评论表的字段结构",
expected_artifacts=["docs/schema.md"],
writable_files=["docs/schema.md"]
)
print(f" [OK] Task 1: {task1.name}")
# Task 2: blockedBy Task 1
task2 = queue.create_task(
name="实现后端 API",
description="实现 CRUD 接口",
blocked_by=[task1.id],
expected_artifacts=["src/api/users.py", "src/api/posts.py"],
writable_files=["src/api/"],
read_only_files=["docs/schema.md"]
)
print(f" [OK] Task 2: {task2.name} (blockedBy: {task1.id})")
# Task 3: blockedBy Task 2
task3 = queue.create_task(
name="实现前端页面",
description="实现用户列表和文章页面",
blocked_by=[task2.id],
expected_artifacts=["src/frontend/index.html"],
writable_files=["src/frontend/"],
read_only_files=["src/api/"]
)
print(f" [OK] Task 3: {task3.name} (blockedBy: {task2.id})")
# Task 4: blockedBy Task 2 (与 Task 3 并行)
task4 = queue.create_task(
name="编写测试",
description="为 API 编写单元测试",
blocked_by=[task2.id],
expected_artifacts=["tests/test_api.py"],
writable_files=["tests/"],
read_only_files=["src/api/"]
)
print(f" [OK] Task 4: {task4.name} (blockedBy: {task2.id})")
# Task 5: blockedBy Task 3 和 Task 4
task5 = queue.create_task(
name="集成测试",
description="端到端测试",
blocked_by=[task3.id, task4.id],
expected_artifacts=["tests/e2e_test.py"],
writable_files=["tests/"]
)
print(f" [OK] Task 5: {task5.name} (blockedBy: {task3.id}, {task4.id})")
print("\n" + "=" * 60)
print("任务依赖图 (DAG)")
print("=" * 60)
print("""
[Task 1: 设计 Schema]
|
V
[Task 2: 实现后端 API]
|
+-----+-----+
V V
[Task 3: 前端] [Task 4: 测试]
| |
+-----+-----+
V
[Task 5: 集成测试]
""")
# 显示所有任务
print("\n" + "=" * 60)
print("所有任务")
print("=" * 60)
for task in queue.list_tasks():
print_task(task)
# 测试可用任务计算
print("\n" + "=" * 60)
print("可用任务 (无依赖或依赖已满足)")
print("=" * 60)
available = queue.get_available_tasks()
for task in available:
print(f" -> {task.name}")
# 测试任务认领
print("\n" + "=" * 60)
print("测试文件锁抢任务")
print("=" * 60)
# Worker 1 尝试认领 Task 1
success1 = queue.claim_task(task1.id, "worker-8001")
print(f" Worker 8001 认领 Task 1: {('[OK]' if success1 else '[FAIL]')}")
# Worker 2 尝试认领 Task 1 (应该失败)
success2 = queue.claim_task(task1.id, "worker-8002")
print(f" Worker 8002 认领 Task 1: {('[OK]' if success2 else '[FAIL] 被抢走了')}")
# Worker 2 尝试认领 Task 2 (应该失败 - 依赖未完成)
success3 = queue.claim_task(task2.id, "worker-8002")
print(f" Worker 8002 认领 Task 2: {('[OK]' if success3 else '[FAIL] Task 1 未完成')}")
# 完成 Task 1
print("\n Worker 8001 完成 Task 1...")
queue.complete_task(task1.id)
print(" Task 1 已标记完成")
# 再次检查可用任务
print("\n" + "=" * 60)
print("Task 1 完成后,可用的任务")
print("=" * 60)
available = queue.get_available_tasks()
for task in available:
print(f" -> {task.name}")
# Worker 2 现在可以认领 Task 2
success4 = queue.claim_task(task2.id, "worker-8002")
print(f"\n Worker 8002 认领 Task 2: {('[OK]' if success4 else '[FAIL]')}")
# 测试 Wave 计算
print("\n" + "=" * 60)
print("执行 Wave 分析")
print("=" * 60)
# 模拟完成所有任务
queue.complete_task(task2.id)
queue.claim_task(task3.id, "worker-8001")
queue.claim_task(task4.id, "worker-8003")
available = queue.get_available_tasks()
print(f" Wave 1 (Task 1, 2 完成):")
print(f" 可用: {[t.name for t in available]}")
queue.complete_task(task3.id)
queue.complete_task(task4.id)
available = queue.get_available_tasks()
print(f" Wave 2 (Task 3, 4 完成):")
print(f" 可用: {[t.name for t in available]}")
# 统计
print("\n" + "=" * 60)
print("最终统计")
print("=" * 60)
stats = queue.get_task_stats()
print(f" 总任务数: {stats['total']}")
print(f" 已完成: {stats['completed']}")
print(f" 进行中: {stats['in_progress']}")
print(f" 等待中: {stats['pending']}")
print("\n" + "=" * 60)
print("DAG 测试完成!")
print("=" * 60)
finally:
# 清理
print(f"\n清理测试目录: {temp_dir}")
shutil.rmtree(temp_dir, ignore_errors=True)
def test_mailbox():
"""测试 Mailbox 通信"""
print("\n" + "=" * 60)
print("测试 Mailbox Agent 间通信")
print("=" * 60)
from mailbox import Mailbox, Message
temp_dir = tempfile.mkdtemp(prefix="mailbox_test_")
try:
mb = Mailbox(base_dir=temp_dir)
# Worker 8001 发送消息给 Lead
msg_id = mb.send_message(
from_agent="worker-8001",
to_agent="lead",
content="Task 1 已完成!",
msg_type="task_completed",
metadata={"task_id": "task-1", "artifacts": ["docs/schema.md"]}
)
print(f" [OK] Worker 8001 -> Lead: 消息 ID = {msg_id}")
# Lead 发送消息给 Worker 8002
msg_id2 = mb.send_message(
from_agent="lead",
to_agent="worker-8002",
content="开始 Task 2",
msg_type="task_assignment"
)
print(f" [OK] Lead -> Worker 8002: 消息 ID = {msg_id2}")
# Worker 8002 检查收件箱
messages = mb.read_messages("worker-8002")
print(f" Worker 8002 收件箱: {len(messages)} 条消息")
for msg in messages:
print(f" - [{msg.msg_type}] {msg.content}")
# Lead 检查收件箱
messages = mb.read_messages("lead")
print(f" Lead 收件箱: {len(messages)} 条消息")
# 广播测试
print("\n 测试广播...")
mb.broadcast(
from_agent="lead",
content="所有 Worker 停止工作!",
agent_ids=["worker-8001", "worker-8002", "worker-8003"]
)
print(" [OK] 广播已发送")
print("\nMailbox 测试完成!")
finally:
shutil.rmtree(temp_dir, ignore_errors=True)
def test_path_guard():
"""测试路径守卫"""
print("\n" + "=" * 60)
print("测试 PathGuard 路径守卫")
print("=" * 60)
from path_guard import PathGuard
# 模拟项目目录
project_dir = "D:\\projects\\my_blog"
# Agent 目录 (禁止 Worker 访问)
agent_dir = "D:\\git_codes\\google_adk_agent\\skills"
guard = PathGuard(
allowed_root=project_dir,
forbidden_paths=[agent_dir]
)
# 测试允许的路径
test_cases = [
("D:\\projects\\my_blog\\src\\api\\users.py", True, "项目内文件"),
("D:\\projects\\my_blog\\docs\\schema.md", True, "项目内文档"),
("D:\\git_codes\\google_adk_agent\\skills\\agent_team\\tools.py", False, "Agent 目录 (禁止)"),
("D:\\windows\\system32\\config.sys", False, "系统目录 (禁止)"),
]
for path, expected, description in test_cases:
result = guard.is_allowed(path)
status = "[OK]" if result == expected else "[FAIL]"
print(f" {status} {description}: {result}")
print("\nPathGuard 测试完成!")
if __name__ == "__main__":
print("\n" + "=" * 60)
print("Agent Team DAG 系统测试")
print("=" * 60 + "\n")
# 测试 DAG 功能
test_basic_dag()
# 测试 Mailbox
test_mailbox()
# 测试 PathGuard
test_path_guard()
print("\n" + "=" * 60)
print("所有测试完成!")
print("=" * 60)
Related skills
| Skill | Stars | Updated | Works with | Status | Install |
|---|---|---|---|---|---|
| reverse-engineer-anythingmorluto/reaReverse engineer native, managed, Electron/JavaScript, packaged, firmware, and browser targets with REA, and build agent environments (AgentEnv bundles) of web applications from a live instance or a capture, and of Linux desktop apps with a task from a request. Use shipped-artifact or requested runtime evidence to explain features, compare versions, decompile code, or guide a reconstruction. Skip REA for ordinary source-repository architecture analysis. | Stars: 469k | Updated | Claude CodeCodex | Verified | Install |
| diagnosing-superpowersobra/superpowersUse when a superpowers session went wrong and your human partner wants to know why — repeated work, ignored plans, stumbles, poor results, a skill that didn't fire, "it took too long", "why is it so expensive", "what is it doing" — or wants to build a bug report for the superpowers maintainers, for the current session or a past one identified by id or path, on any harness. | Stars: 321k | Updated | Claude CodeCodex | Verified | Install |
| dispatching-parallel-agentsobra/superpowersUse when facing 2+ independent tasks that can be worked on without shared state or sequential dependencies | Stars: 321k | Updated | Claude CodeCodex | Verified | Install |
| executing-plansobra/superpowersUse when executing an implementation plan in the current session as the implementer yourself — your human partner chose inline execution, or no subagent tool is available | Stars: 321k | Updated | Claude CodeCodex | Verified | Install |