What problem does it solve?
AI-assisted code changes often introduce repeatable regressions because the same model writes and reviews fixes, causing blind spots such as mismatched sandbox and production paths, missing SELECT fields, and incorrect error-state handling. This Skill provides focused strategies and test patterns to detect and prevent those regressions early in the development workflow.
Core Features & Use Cases
- Sandbox-first API Tests: Guidance to force sandbox mode and run lightweight API tests without a database to validate response shapes.
- Regression-by-Example: Emphasizes writing tests for observed bugs (not hypothetical cases) so each fixed issue gains a targeted test that prevents recurrence.
- Workflow Integration: Defines a bug-check sequence combining automated tests, build/type checks, and AI-assisted review to ensure fixes include regression tests.
- Use Case: When a new API field is added by an AI patch, run sandbox tests and a bug-check to ensure the field appears in sandbox and production responses and is included in DB selects.
Quick Start
Run the test suite and build checks, then invoke the bug-check workflow to have the system run sandbox API tests, report failures, and propose regression tests for any confirmed bugs.