architect-error-handling

What problem does it solve?

Many implementations treat error handling as an afterthought, producing inconsistent logs, unclear caller behavior, unsafe retries, and partial failures that leave systems in unknown states; this Skill helps design error handling up front so failures are detected, contained, observable, and recoverable.

Core Features & Use Cases

• Error boundary decisions: Guidance on where to catch, translate, or let errors propagate across controllers, services, and infrastructure.
• State and recovery strategies: Patterns for atomic operations, idempotency keys, compensating actions, sagas, and documenting partial-failure recovery.
• Caller and operator contracts: Recommendations for API error formats, retry categorization (retryable vs non-retryable), logging/alerting context, and a PR attestation template to communicate decisions.
• Use Case: Designing error handling for a Create Order flow that involves payment, inventory reservation, and shipment creation so that failures either roll back safely or allow systematic recovery.

Quick Start

Create an error handling summary for the Create Order flow that lists where to catch errors, the state guarantees (atomic/idempotent/compensating), retry rules, and a short attestation for the PR.