workflow-validate

Run this Skill after completing an analysis document. It selects an appropriate validation method such as prototyping, TDD, or library exploration, executes it, and records a clear success or failure verdict with next steps in a validation report.

It supports six methods: prototype implementation for core logic, Playwright verification for UI changes, TDD for complex algorithms, library exploration for external dependencies, documentation research for standards, and user delegation for environment-specific checks.

Use a prototype when core logic feasibility is uncertain and you need working code as proof. Use TDD when validating complex algorithms where edge cases and correctness matter more than a runnable demonstration.

A failure verdict means the core approach is not feasible or has blocking issues without workarounds. The Skill directs you back to the analysis phase to re-evaluate alternatives rather than proceeding to planning.

Prototype code is stored in a __prototype__/{feature-name}/ directory. It serves as a reference for later plan and execute phases, and the user manages its cleanup since prototypes are throwaway learning artifacts.