What problem does it solve? Turning vague feature ideas into unambiguous, testable specifications is hard: requirements get missed, designs accumulate unjustified complexity, and implementation sessions lack the context to proceed independently. This Skill provides a structured process that produces a numbered-requirement spec a fresh session can implement from directly. ## Core Features & Use Cases - Structured five-step process: codebase analysis, domain research, discovery interview, draft-and-red-team loop, and final spec writing with nested-numbered requirements. - Adversarial refinement: a red-team checklist stress-tests design proportionality, edge cases, security, and operability, with each pass required to reduce or hold complexity. - Adaptive compression: skip rules for trivial changes, greenfield work, refactors, and prototypes, while never skipping discovery or numbered requirements. - Use Case: A developer planning a new API feature uses this Skill to interview stakeholders, red-team the architecture across multiple cycles, and produce a self-contained markdown spec with testable requirements like 2.14 that another engineer or AI session can implement without conversation history. ## Quick Start Ask the AI to help you write a spec for the feature you have in mind, starting with an analysis of the relevant codebase and a discovery interview.
