What problem does it solve? Non-trivial code changes often ship with unvalidated inputs, swallowed errors, unbounded retries, and hidden dependencies that only surface as production incidents. This Skill turns an implementation plan into code where the safe path is the natural path, catching structural defects while the code is being written rather than in review or on-call. ## Core Features & Use Cases - Boundary enforcement: Parse and validate input at the door into narrower types, bound caller-controlled growth, and keep raw DTOs out of core logic. - Runtime safety checks: Enforce deadlines on cross-boundary waits, atomic or serialized shared-state mutation, idempotent retries, and resource release on every exit path. - Operability while building: Make failures visible in the language idiom, emit signals alongside the code, and test contracts, parsers, failures, and replay as you go. - Use Case: While implementing a new endpoint handler that writes to a queue, use this Skill to verify the input is schema-parsed, the mutation is idempotent against duplicate clicks, the queue client is behind a testable seam, and errors are never silently caught. ## Quick Start Ask the AI to apply structural implementation checks while writing a new endpoint handler, background job, or migration so parsing, idempotency, deadlines, and failure visibility are built in from the start.
