← Back to Skills Marketplace
broommonk

Async Programming

by broommonk · GitHub ↗ · v1.0.0
cross-platform ⚠ suspicious
645
Downloads
0
Stars
2
Active Installs
1
Versions
Install in OpenClaw
/install async-programming
Description
异步编程任务处理。当用户请求编程任务时,立即调用子 agent + 立即回复确认,无需等待完成即可继续聊天。
README (SKILL.md)

异步编程任务处理技能

触发条件

当用户请求涉及以下内容时,激活此技能:

  • ✅ 编写/修改代码
  • ✅ 添加新功能
  • ✅ 修复 bug
  • ✅ 重构代码
  • ✅ 创建脚本
  • ✅ 项目配置
  • ✅ 依赖安装

核心原则

收到编程任务 → 立即调用子 agent + 立即回复确认 = 可以继续聊天


执行流程

1. 调用子 agent

sessions_spawn({
  task: "{具体任务描述,包含项目路径和具体要求}",
  label: "{项目名} - {任务类型}",
  model: "bailian/qwen3-coder-plus",
  mode: "run",
  runTimeoutSeconds: 900
})

参数说明:

参数 说明 示例
task 明确的任务描述 "给 Stock-Analysis 添加 Excel 导出功能"
label 任务标识,格式 项目名 - 任务类型 Stock-Analysis - 功能开发
model 编程任务统一用 coder 模型 bailian/qwen3-coder-plus
runTimeoutSeconds 超时时间(秒) 900 (15 分钟)

2. 立即回复确认(关键!)

调用子 agent 后,立即回复用户,不要等待完成!

回复模板:

✅ 任务已接收

**正在处理:**
- 项目:{项目名}
- 任务:{任务简述}
- 执行者:子 agent(编程助手)
- 预计时间:2-5 分钟

你可以在等待期间继续问我其他问题,完成后我会通知你。

3. 完成后汇报

子 agent 完成后会自动发送结果,转发给用户:

✅ {任务名} 完成了!

**完成内容:**
- {改动 1}
- {改动 2}

**文件变更:**
- {文件路径 1}
- {文件路径 2}

需要调整或继续开发吗?

多任务并发

用户可以连续下达多个编程任务,子 agent 并行执行:

用户:"给项目 A 加个功能"
  → 调用子 agent #1
  → 立即回复确认

用户:"帮项目 B 修复 bug"
  → 调用子 agent #2
  → 立即回复确认

用户:"项目 C 也帮我看看"
  → 调用子 agent #3
  → 立即回复确认

并发限制: 最多 8 个子 agent 同时运行(agents.defaults.subagents.maxConcurrent


任务追踪

当用户询问进度时,使用:

subagents(action="list")

回复格式:

**当前任务进度:**

✅ 已完成 (2)
- Stock-Analysis - 功能开发
- Amazon-Analyzer - 数据抓取

⏳ 进行中 (1)
- Skills-Bank - 文档整理 (预计还需 1 分钟)

❌ 失败 (0)

示例对话

示例 1:单个任务

用户:给 Stock-Analysis 加个导出 Excel 的功能

助手:
✅ 任务已接收

**正在处理:**
- 项目:Stock-Analysis
- 任务:添加 Excel 导出功能
- 执行者:子 agent(编程助手)
- 预计时间:2-5 分钟

你可以在等待期间继续问我其他问题,完成后我会通知你。

---(2 分钟后,子 agent 完成)---

✅ Excel 导出功能完成了!

**完成内容:**
- 新增 `export_to_excel()` 函数
- 支持导出 CSV 和 JSON 结果
- 添加 openpyxl 依赖

**文件变更:**
- /root/Stock-Analysis/stock_analyzer.py (新增 35 行)
- /root/Stock-Analysis/requirements.txt (新增 openpyxl)

需要测试或继续开发吗?

示例 2:多任务并发

用户:帮我给三个项目都加个 README

助手:
✅ 收到!正在并行处理三个项目:

**任务队列:**
1. Stock-Analysis - 完善 README
2. Amazon-Analyzer - 完善 README
3. Skills-Bank - 完善 README

预计 5-8 分钟完成,你可以继续问我其他问题。

---(5 分钟后)---

✅ 三个项目的 README 都完成了!

**完成情况:**
- ✅ Stock-Analysis - 已添加安装和使用说明
- ✅ Amazon-Analyzer - 已添加 API 配置说明
- ✅ Skills-Bank - 已添加技能分类说明

需要调整吗?

注意事项

  1. 不要等待子 agent 完成才回复 - 这是核心原则
  2. 任务描述要清晰 - 子 agent 需要明确的指令
  3. label 要规范 - 方便追踪和管理
  4. 超时处理 - 15 分钟未完成,主动告知用户
  5. 错误处理 - 子 agent 失败时,解释原因并提供解决方案

相关配置

子 agent 配置位于 openclaw.json

{
  "agents": {
    "defaults": {
      "subagents": {
        "maxConcurrent": 8,
        "maxSpawnDepth": 2,
        "model": "bailian/qwen3.5-plus",
        "runTimeoutSeconds": 600,
        "archiveAfterMinutes": 60
      }
    }
  }
}

此技能确保贾维斯可以并行处理多个编程任务,同时保持与用户的即时沟通

Usage Guidance
This skill is coherent for its stated purpose: it will autonomously spawn child agents to do programming work and immediately confirm to the user. Before installing, verify the platform-level permissions and safeguards for sessions_spawn/subagents: confirm what filesystem and network access spawned subagents have, check max concurrent and timeout settings (manifest shows maxConcurrent=8, SKILL.md uses runTimeoutSeconds=900 while defaults show 600), and ensure you are comfortable with agents making code changes to repositories (examples reference /root paths). Also confirm the model used (bailian/qwen3-coder-plus) is acceptable for your environment. If you need stricter limits, ask for explicit sandboxing, reduced concurrency, or review hooks that require manual approval before code-modifying subagents run.
Capability Analysis
Type: OpenClaw Skill Name: async-programming Version: 1.0.0 The `SKILL.md` file instructs the agent to spawn a sub-agent (`sessions_spawn`) for programming tasks. The `task` parameter for this sub-agent is directly populated with user-provided input, and the sub-agent is specified as a 'coder model'. This creates a significant prompt injection vulnerability, allowing a malicious user to potentially instruct the sub-agent to perform unauthorized actions, including arbitrary code execution or data exfiltration, by crafting a harmful task description.
Capability Assessment
Purpose & Capability
Name/description, skill.yaml requirements (sessions_spawn, subagents) and SKILL.md all describe the same capability: spawn subagents to run programming tasks and immediately reply. No unrelated environment variables, binaries, or external endpoints are requested.
Instruction Scope
SKILL.md explicitly instructs using platform APIs (sessions_spawn, subagents(action="list")) and to supply task, label, model, timeout. This matches the purpose. Note: the skill relies on subagents executing code against project paths (examples reference /root/...), but it does not define access controls or limits on what spawned subagents may read/write. Also the SKILL.md prescribes a specific coder model (bailian/qwen3-coder-plus). These are expected for the goal but you should confirm platform-level safeguards for subagent filesystem and network access.
Install Mechanism
Instruction-only skill with no install spec and no code files to write/execute. Low installation risk.
Credentials
No environment variables, credentials, or config paths are required beyond referencing openclaw.json defaults. The declared requirements (sessions_spawn, subagents) are proportional to the stated functionality.
Persistence & Privilege
always:false and default autonomous invocation are used (normal). The skill spawns subagents but does not request persistent system-wide privileges or modify other skills. Consider that autonomous subagent spawning is the intended behavior and grants runtime capability to run code.
How to Use
  1. Make sure OpenClaw is installed (local or Docker)
  2. Run the install command in chat: /install async-programming
  3. After installation, invoke the skill by name or use /async-programming
  4. Provide required inputs per the skill's parameter spec and get structured output
Version History
v1.0.0
async-programming v1.0.0 - Initial release: enables asynchronous handling of coding tasks. - Instantly calls a sub agent upon user’s task request and immediately confirms receipt—no need to wait for task completion before chatting further. - Supports multiple concurrent programming tasks (up to 8 in parallel). - Provides clear progress updates and notifications when tasks are completed or fail. - Includes templates for task confirmation, completion reports, and progress tracking.
Metadata
Slug async-programming
Version 1.0.0
License
All-time Installs 2
Active Installs 2
Total Versions 1
Frequently Asked Questions

What is Async Programming?

异步编程任务处理。当用户请求编程任务时,立即调用子 agent + 立即回复确认,无需等待完成即可继续聊天。 It is an AI Agent Skill for Claude Code / OpenClaw, with 645 downloads so far.

How do I install Async Programming?

Run "/install async-programming" in the OpenClaw or Claude Code chat to install it in one step — no extra setup required.

Is Async Programming free?

Yes, Async Programming is completely free (open-source). You can download, install and use it at no cost.

Which platforms does Async Programming support?

Async Programming is cross-platform and runs anywhere OpenClaw / Claude Code is available (cross-platform).

Who created Async Programming?

It is built and maintained by broommonk (@broommonk); the current version is v1.0.0.

💬 Comments