Task Executor

任务执行器。接收用户需求，自动拆分任务，异步执行，返回结果。 **核心功能**： - 需求输入：用户直接说需求 - 任务拆分：AI 分析，自动拆分 - 自动执行：subagent 异步执行 - 进度跟踪：本地状态 **使用场景**： - "帮我分析AI发展趋势" - "调研某个话题" - "写一份报告

Security Scan

The skill's declared purpose (automatically splitting and running sub‑tasks) aligns with its instructions and required resources; it is an instruction-only orchestration recipe that does not request unexpected credentials or install code.

Purpose & Capability

Name and description describe a multi-agent task executor; the SKILL.md contains only orchestration instructions (task split, search, document, analysis, async execution). There are no declared env vars, binaries, or install steps that would be unrelated to the stated purpose.

Instruction Scope

Instructions are high-level orchestration guidance and mention spawning subagents (sessions_spawn), parallel execution, and storing simple local state (memory/tasks/active.json). This scope is consistent with the skill's purpose, but the instructions are vague about which external search endpoints or agents to call and how subagents are authenticated. The doc also mentions an optional Feishu (Lark) spreadsheet integration but does not declare required credentials.

Install Mechanism

No install spec and no code files — the skill is instruction-only, so nothing is written to disk by an installer and there are no third‑party packages to review.

Credentials

The skill declares no required environment variables or credentials, which is proportional. One caveat: it suggests optional Feishu table sync (which would require Feishu credentials if enabled) but does not declare or require those creds in the metadata; users should expect to supply such credentials only if they opt into that integration.

Persistence & Privilege

always is false and autonomous invocation is allowed (platform default). The skill proposes writing a local state file under memory/tasks/active.json — this is reasonable for progress tracking and is limited in scope. The skill does not request system‑wide modifications or other skills' configs.

Assessment

This skill is an instruction-only orchestration recipe for splitting and running sub‑tasks; it does not require credentials or install anything. Before installing: be aware it will ask the agent to spawn subagents and perform web searches, and it proposes writing a local state file (memory/tasks/active.json). If you enable the optional Feishu integration you will need to provide Feishu credentials — those are not included in the skill metadata, so only supply them if you trust the skill and understand where data will be sent. Because the instructions are high-level, monitor the first runs to see what external endpoints or prompts for credentials the agent produces and limit access to sensitive files or secrets during testing.

任务执行器 Skill (v4.0.0)

🎯 核心价值

用户给一个需求 → AI 自动完成，不需要跟踪细节

---

⚡ 核心流程

用户需求
    ↓
AI 拆分任务
    ↓
并行执行
    ↓
返回结果

---

📋 Agent 角色

Agent	职责
主Agent	接收需求、拆分任务、协调、执行
搜索Agent	网络搜索、信息收集
文档Agent	撰写文档、编辑内容
分析Agent	数据分析、处理

---

🎯 核心功能

1. 接收需求

用户：帮我分析AI发展趋势

2. 拆分任务

AI 分析需求，拆分成可执行的任务：

- 搜索AI最新动态（搜索Agent）
- 整理行业报告（分析Agent）
- 撰写分析报告（文档Agent）

3. 执行任务

并行执行各个任务：

搜索Agent → 搜索信息
文档Agent → 创建文档
分析Agent → 处理数据

4. 返回结果

完成后通知用户：

✅ 分析完成！
文档链接：xxx

---

📖 使用示例

例1: 简单需求

用户：帮我搜索量子计算最新进展

主Agent：
1. 拆分任务：搜索信息
2. 执行：调用搜索
3. 返回结果

例2: 复杂需求

用户：帮我分析新能源汽车行业

主Agent：
1. 拆分任务：
   - 搜索行业数据（搜索Agent）
   - 整理市场分析（分析Agent）
   - 撰写报告（文档Agent）
2. 异步执行任务
3. 汇总结果
4. 通知用户

---

📁 状态管理（可选）

本地状态文件

memory/tasks/active.json

{
  "tasks": [
    {
      "id": "task-001",
      "需求": "分析AI趋势",
      "状态": "已完成",
      "结果": "文档链接"
    }
  ]
}

飞书表格（可选扩展）

用户需要可视化时，可以添加飞书表格同步。

---

🔗 触发关键词

帮我、分析、调研、写报告
创建需求、执行任务

---

⚠️ 注意事项

1. 异步执行：使用 sessions_spawn 不阻塞
2. 并行执行：多个任务同时进行
3. 结果汇总：完成后统一通知用户
4. 状态可选：可以不用飞书表格

---

版本: 4.0.0 | 2026-03-22