Async Python Patterns

Use when building or reviewing Python asyncio services, concurrent I/O, async APIs, background tasks, cancellation, or event-loop performance.

9 stars 3 forksSoftware Development

Install

skills CLI (Claude Code, Cursor, Codex and more)
npx skills add sandbaseai/workbuddy-skill --skill async-python-patterns
Manual install
git clone --depth 1 https://github.com/sandbaseai/workbuddy-skill.git /tmp/workbuddy-skill
mkdir -p ~/.claude/skills
cp -r /tmp/workbuddy-skill/skills/async-python-patterns ~/.claude/skills/async-python-patterns

Personal skills live in ~/.claude/skills. Use .claude/skills in a project folder to share a skill with your team.

From the SKILL.md

# Async Python Patterns

用 `asyncio` 管理 I/O 并发,同时保持事件循环可响应、资源可回收、取消可传播、并发有上限。异步不是越多越好:CPU 密集任务、低并发脚本或阻塞 SDK 应选择同步、线程池或进程池,并用基线证明收益。

## 使用边界

- 开始前确认 Python/框架/依赖版本、调用链、并发量、deadline、资源配额、owner、SLO 和目标环境。
- 默认只读检查 coroutine、task、连接池、队列、超时、trace 和错误;在本地/预发布使用合成负载与受控故障。
- 不擅自修改生产并发、超时、线程/进程池、连接池、限流、流量或重启服务;这些变更需要授权和回滚方案。
- 日志、task 参数、异常和 trace 不得包含 token、完整 payload、个人信息或原始响应;使用脱敏摘要和受控 correlation ID。

## 同步/异步边界

多网络/数据库调用适合 asyncio;CPU 密集工作使用 `asyncio.to_thread`(阻塞但可释放事件循环)或进程池,并限制 worker。一个调用路径应有清晰的 sync/async 边界,不能在事件循环中执行 `time.sleep`、同步 HTTP/数据库客户端、重型序列化或无界文件操作。

每个外部调用都要有连接、读和总 deadline;每个 task 都说明 owner、结果、异常、取消和资源清理语义。优先使用 `async with`、`async for` 和结构化并发(如 `TaskGroup`,按 Python 版本确认行为)。

```python
import asyncio

async def fetch_bounded(items, client, *, limit=20):
    semaphore = asyncio.Semaphore(limit)

    async def one(item):
        async with semaphore:
            # The caller supplies a request deadline; no unbounded await.
            return await client.fetch(item, timeout=5.0)

    # Keep the result order and make the concurrency bound explicit.
    return await asyncio.gather(*(one(item) for item in items))
```

## 并发、背压与结果

`gather`、task 集合、队列和连接池都必须有上限;不要一次为不受控输入创建百万个 task。生产者速度超过消费者时使用 `asyncio.Queue(maxsize=...)`、信号量或流式 `async for`,队列满时暂停、丢弃低优先级或快速失败,并返回明确状态。

决定失败策略:默认一个失败是否取消同组任务、是否收集部分结果、是否允许继续。不要用 `return_exceptions=True` 后静默过滤异常;逐项分类、记录受控上下文并为部分成功定义业务语义。对外部调用结合限流、指数退避+jitter、幂等键和总 deadline,避免并发重试放大。

## 取消、超时与清理
Read the full skill on GitHub

Details

Skill name
async-python-patterns
Path
skills/async-python-patterns/SKILL.md
License
MIT
Repo updated
8 September 2026

FAQ

What does the Async Python Patterns skill do?

Use when building or reviewing Python asyncio services, concurrent I/O, async APIs, background tasks, cancellation, or event-loop performance.

How do I install the Async Python Patterns skill?

skills CLI (Claude Code, Cursor, Codex and more): npx skills add sandbaseai/workbuddy-skill --skill async-python-patterns. Skills go in ~/.claude/skills (personal) or .claude/skills (project), and Claude loads them automatically when a task matches the description.

Is Async Python Patterns popular?

It lives in sandbaseai/workbuddy-skill, which has 9 GitHub stars and 3 forks.

Related skills

Browse the most popular Claude skills, all software development skills, or write your own with the skills generator.