What problem does it solve? Applications fail in production due to network timeouts, rate limits, and server errors, and without structured recovery logic these failures cascade into outages. This Skill provides concrete patterns for deciding when to retry, when to fall back, and how to roll back safely. ## Core Features & Use Cases - Retry Decision Rules: Distinguishes retryable failures (timeouts, 429, 5xx) from non-retryable ones (400, 401, 404, business logic errors) to avoid wasted attempts. - Backoff and Circuit Breakers: Provides exponential backoff with jitter formulas and the CLOSED/OPEN/HALF-OPEN circuit breaker state machine. - Fallback and Rollback Patterns: Covers default values, cached responses, degraded service modes, database transactions, sagas, and feature flags. - Use Case: When an API call to a payment service starts timing out, apply the retry-with-backoff pattern for transient failures, trip a circuit breaker after repeated failures, and serve a cached fallback response instead of crashing the checkout flow. ## Quick Start Ask the agent to load the error-recovery-patterns skill and design retry and fallback logic for your failing API integration.