What problem does it solve? Creating a new agent skill from scratch requires dozens of decisions about structure, state management, error handling, and browser automation, and inconsistent hand-rolled skills break in production. This Skill interviews the user, selects the right architecture modules, and generates a complete, production-ready skill directory with no placeholders or TODOs. ## Core Features & Use Cases - Guided Scaffolding Interview: Asks targeted questions about work units, phases, state tracking, authentication, and execution environment to determine which modules the new skill needs. - Dual Generation Paths: Produces either a lightweight knowledge/workflow skill (SKILL.md plus optional references) or a full automation pipeline with Python orchestrator, state store, circuit breaker, and session logging. - Reusable Pattern Catalog: Ships 12 production-tested code patterns covering Google Sheets state stores, JSON state stores, encrypted credential vaults, Playwright browser adapters, preflight validation, and screenshot evidence capture. - Use Case: A user says "create a skill that audits a GitHub repo, discovers features, and screenshots each one" — the scaffolder infers the architecture, then generates config templates, scripts, SKILL.md, setup docs, and a 50+ item QA checklist run. ## Quick Start Ask the agent to create a new skill for the multi-step workflow you want to automate and answer its scoping questions.
