AgentSkillsCN

error-handling

提供具备强大容错能力的应用构建指南，辅以稳健的错误处理策略。在设计 API、排查问题，或实施重试机制、熔断器等韧性模式时使用此方法。

SKILL.md
--- frontmatter
name: error-handling
description: Guide for building resilient applications with robust error handling strategies. Use when designing APIs, debugging issues, or implementing resilience patterns like retries and circuit breakers.

Error Handling Patterns

When to use this skill

  • •Implementing error handling in new features
  • •Designing error-resilient APIs
  • •Debugging production issues or improving application reliability
  • •Creating better error messages
  • •Implementing resilience patterns (retry, circuit breaker)
  • •Handling async/concurrent errors

Workflow

  1. •Analyze Context: Determine the type of error (Recoverable vs Unrecoverable) and the appropriate handling philosophy (Exceptions vs Result Types).
  2. •Select Pattern: Choose the right pattern (e.g., Retry, Circuit Breaker, Error Aggregation) based on the requirement.
  3. •Implement: Apply the pattern using language-specific best practices.
  4. •Verify: Ensure errors are caught, logged, and handled gracefully without crashing the application.

Instructions

1. Philosophies & Categories

  • •Recoverable: Network issues, input validation. Handle with Result types or specific exceptions.
  • •Unrecoverable: OOM, bugs. Let it crash or catch at the top level.
  • •Exceptions: Use for unexpected/exceptional conditions.
  • •Result Types: Use for expected errors (validation, business logic).

2. Implementation by Language

  • •Python: Use custom exception hierarchies (ApplicationError), context managers for cleanup, and decorators for retries.
  • •TS/JS: Use custom Error classes, Result type pattern for explicit handling, and proper async/await try-catch blocks.
  • •Rust: Use Result and Option types.
  • •Go: Use explicit error returns and sentinel errors.

3. Universal Resilience Patterns

  • •Circuit Breaker: For distributed systems to prevent cascading failures.
  • •Error Aggregation: Collect multiple errors (e.g., validation) before failing.
  • •Graceful Degradation: Fallback to cache or default values when primary fails.

4. Best Practices

  • •Fail Fast: Validate early.
  • •Context: Include metadata in errors.
  • •Log properly: Don't spam logs for expected failures.
  • •Clean up: Always release resources (finally, defer).

Resources