on-write-error-handle

Provides error classification by severity/recoverability, recovery strategies, logging, and handlers for auth/rate-limit/timeout errors in resilient services.

Guides error handling best practices to prevent silent failures, preserve context, and log effectively in try-catch blocks, propagation, and Result patterns.

Provides principles and patterns for production AI products: handling LLM variance, prompt engineering as code, RAG preference, latency/cost optimization, structured output with validation, streaming.