What problem does it solve? Users rarely know which built-in skills exist, so prompts go unassisted and quality guardrails never fire. This Skill reads each prompt before acting, silently on clean prompts, and surfaces the right skill, closes ambiguity, or fires always-on guardrails for design, strategy, source lookup, spec governance, operator questions, and workspace path resolution. ## Core Features & Use Cases - Lean prompt triage: Stays silent on clean prompts and only pauses when a built-in skill matches the job or the request is genuinely ambiguous, offering up to three labeled moves with a marked recommendation. - Six always-on guardrails: Automatically fires advisor selection for design and strategy work, Source-First answering from existing specs or records, Spec-Map lookup before building in governed domains, an Ask Gate that blocks re-asking already-ruled questions, and a Door Gate that resolves registered workspace roots before reporting files unreachable. - Three user modes: "Only when it helps" (default), "Always 3 options", and "Only when I ask", so the user controls how often the triage layer pauses. - Use Case: A user asks to build a landing page. The design guardrail fires automatically, resolving the company's brand assets by category and loading the right advisors before any design work begins. ## Quick Start Ask the assistant to review your next prompt and suggest whether a built-in skill fits before it starts working.
