What problem does it solve? Vague goals like "help me run my business with AI" often lead to over-engineered platforms or one-off prompts that never become repeatable workflows. This Skill turns a messy idea into a confirmed work target and then into a runnable AI work system with explicit human-approval gates, verification checks, and a small trial run. ## Core Features & Use Cases - Two-layer design process: Layer 1 refines a vague idea into a confirmed work target through an Idea Diagnostic and Confirmation Gate; Layer 2 produces a Work System Card with separate Prompt, Context, Harness, and Loop layers. - Right-sized architecture decisions: Decides whether the job needs a prompt, context pack, harness, loop, MCP, orchestrator, or reusable skill, treating MCPs and orchestrators as outputs rather than prerequisites. - Safety and verification built in: Enforces permission tiers so a minimal first version never performs external real-world writes, applies a verification ladder, and runs a maker-checker review gate before handoff. - Use Case: A founder pastes messy weekly facts and asks for an AI operating system; the Skill returns a plain-language brief, a weekly decision-review loop with human approval gates, a copyable starting prompt, and a trial run instead of a platform. ## Quick Start Use the system-wright skill to turn my vague goal of managing weekly business decisions with AI into a reusable work system with a trial run.
