error-handling_skill

Overview

This skill teaches practical error handling strategies to improve reliability and observability in multi-agent systems. It focuses on validation, recovery mechanisms, and clear feedback so failures are managed predictably. The goal is to maintain system stability while giving developers and users actionable information.

How this skill works

The skill inspects code paths for likely failure points, enforces input validation, and applies structured exception handling or error returns. It integrates logging, contextual error messages, and retry/fallback logic for transient faults. For critical failures it recommends fail-fast behavior and for external errors it prescribes graceful recovery and clear user-facing feedback.

When to use it

• When writing any code that can fail, including agents and background workers
• When calling external APIs or services with potential timeouts or errors
• During user input processing and validation paths
• When performing file I/O, configuration loading, or parsing
• For network requests, database access, and inter-process communication

Best practices

• Identify and document failure modes early in design
• Validate inputs and fail fast for programming errors
• Log errors with context (operation, inputs, correlation IDs) without leaking secrets
• Implement retries with backoff for transient failures and circuit breakers for persistent issues
• Return clear, actionable messages to callers; escalate unexpected errors

Example use cases

• Wrap configuration loading with specific handlers for missing files, parse errors, and permission issues
• Add retry logic and exponential backoff when interacting with flaky external APIs
• Validate agent task payloads at the queue boundary and reject malformed tasks with reasons
• Log and surface errors in automated workflows so operators can triage failures quickly
• Use typed error classes to distinguish recoverable vs. fatal failures in worker processes