tenacity-retry

Implement retry logic for Python API calls by applying Tenacity decorators to your functions. This automatically handles transient failures using exponential backoff with jitter, simplifying integration for unreliable I/O operations.

Yes, you can use retry decorators with async Python functions. The Skill supports both synchronous and asynchronous functions, allowing you to apply exponential backoff and jitter to concurrent network operations.

The best way to retry LLM calls and database operations in Python is using prebuilt Tenacity decorators. The Skill provides ready-made patterns like retry_llm_call and retry_db_operation to quickly integrate resilient behavior.

Exponential backoff with jitter for network requests works by progressively increasing the wait time between retries and adding random variance. This prevents thundering herd problems when retrying transient network failures.

You do not strictly need to configure custom retry rules for specific error patterns, as prebuilt utilities exist. However, a configurable retry factory is available to generate custom configurations for unique error scenarios.