Error Handling

Choose error mechanisms by recoverability and caller responsibility.

Workflow

1. Classify each error as domain, validation, application, infrastructure, programming defect, or fatal system condition.

2. Use typed Result/Either-style returns for recoverable domain and validation errors.

3. Use exceptions or panics for defects and truly exceptional infrastructure failures when callers cannot make a local decision.

4. Preserve causal context without leaking secrets.

5. Put retry, timeout, and circuit-breaker policy at integration boundaries.

Rules

• Domain errors should be explicit in function signatures where the language supports it.