菜单
Async resilience, retries, and HTTP client patterns. Triggers on: retry, timeout, HTTP request, rate limit, aiohttp, ClientSession, backoff.
用 Codex 或 Claude 帮你安装 复制这段 Prompt，粘贴到 Codex、Claude 或其他助手里，让它检查 Skill 页面并帮你完成安装。
基于 SOC 职业分类
| name | python-resilience-patterns |
| description | Async resilience, retries, and HTTP client patterns. Triggers on: retry, timeout, HTTP request, rate limit, aiohttp, ClientSession, backoff. |
while retry loops with asyncio.sleep().src.utils.retry.retry_with_backoff + RetryConfig — the canonical retry utility in this project.aiohttp.ClientSession as an async context manager. Never leave sessions unclosed.aiohttp.ClientTimeout explicitly — no bare session.get(url) without a timeout.import aiohttp
from src.utils.retry import RetryConfig, retry_with_backoff
@retry_with_backoff(
config=RetryConfig(max_attempts=3, base_delay=2.0, max_delay=30.0),
exceptions=(aiohttp.ClientError, TimeoutError),
)
async def fetch_platform_data(url: str) -> dict:
async with aiohttp.ClientSession(
timeout=aiohttp.ClientTimeout(total=10.0)
) as session:
async with session.get(url) as response:
response.raise_for_status()
return await response.json()
RetryConfig defaults: max_attempts=3, base_delay=1.0, max_delay=60.0, jitter=True.
Share one session per use case / adapter call, not one per request.
async with aiohttp.ClientSession(
timeout=aiohttp.ClientTimeout(total=30.0)
) as session:
data = await fetch_platform_data(session, url)
details = await fetch_video_details(session, video_id)
retry_async is available for ad-hoc retries without decoration.
from src.utils.retry import RetryConfig, retry_async
result = await retry_async(
some_async_func,
arg1,
arg2,
config=RetryConfig(max_attempts=5, base_delay=1.0),
exceptions=(aiohttp.ClientError,),
)
| ❌ Anti-pattern | ✅ Correct |
|---|---|
while retries > 0: await asyncio.sleep(5) | @retry_with_backoff(config=RetryConfig(...)) |
aiohttp.ClientSession() outside context manager | async with aiohttp.ClientSession() as session: |
session.get(url) without timeout | aiohttp.ClientTimeout(total=10.0) on session |
| Importing tenacity directly | Use src.utils.retry — the project's canonical utility |
Hexagonal Architecture (Ports & Adapters) for the video publishing pipeline. Triggers on: VideoPublisher, TrendingVideoFetcher, TimeSeriesReader, VideoMetadataReader, AuthCredentialStore, ReleaseDateValidator, OAuthProvider, publisher_registry, ports, adapters, domain, hexagonal, build_publishers, TaskGroup publish.
Type hint patterns for this repo: Protocol, TypeVar, TypedDict, Literal, Final, and strict ty usage. Triggers on: type hints, typing, TypeVar, Protocol, Generic, TypedDict, ty check.
Design, review, and optimize Dockerfiles, Docker images, and Docker Compose configurations for production, performance, and maintainability. Use when auditing Docker configurations, creating production-ready images, designing multi-container architectures, or implementing container orchestration with focus on security, reproducibility, minimal attack surface, build optimization, and observability.
Best practices for modeling, ingesting, querying, and analyzing time-series data with TinyFlux. Use when designing TinyFlux Point schemas, implementing query patterns, managing retention and partitioning, or integrating TinyFlux with pandas/NumPy for analysis and forecasting workflows.
Video processing migration (C1 split): guidelines for decomposing src/video_processing.py into src/infrastructure/video/. Triggers on: video_processing, VideoAssetManager, VideoRenderer, ThumbnailGenerator, VideoCompositor, moviepy, CompositeVideoClip, PIL Image.
Structured logging with structlog. Triggers on: log, logger, exception, structlog, observe, print, logging.