Python Async Best Practices

Guidelines for writing and reviewing asyncio code. 5 rules in 1 category, prioritized by impact.

A rule match is a signal, not a verdict. These failures typically pass single-request smoke tests and surface under load — weigh the rule against the code's real concurrency profile.

Quick-reference lines are triggers, not licenses: before applying a rule as a review finding or a transformation, open the rule file and check its counter-signal — the marker-opened paragraph (**When ...** / **Scope:** / **Keep ...**) saying when NOT to apply it.

When to Apply

• Writing or reviewing async def code, task spawning, or streaming consumers
• Auditing an async service for blocking calls or unbounded fan-out
• Debugging hangs, orphaned tasks, or unraisable warnings at teardown

Impact Levels

• HIGH — stalls or silent failures affecting every task on the loop. Fix when found.
• MEDIUM — resource and cleanup discipline; apply to new code and code under review.