What problem does it solve? Turning a vague wish like "automate part of my job" into a working, tested AI system is hard: people skip scoping, over-engineer, or deploy untested automations. This Skill provides a structured Embed → Scope → Build → Prove process that converts a real work problem into a running agentic setup using MimiWork's own primitives. ## Core Features & Use Cases - Structured scoping interview: Embeds in the user's actual workflow, inspects real folders before designing, and writes a 3-line scope with a measurable success metric. - Pattern selection and assembly: Chooses the simplest fitting agentic design pattern (tool use, planning, reflection, ReAct, or multi-agent) and assembles it from scheduled tasks, saved skills, file tools, and office document writers. - Proof before handover: Runs the system once, judges output against the success metric, iterates at most twice, and either deploys with a review date or recommends killing the pilot. - Use Case: A user says "every Monday, summarize the new files in my reports folder into a one-page brief." The Skill scopes the job, builds a scheduled task with reflection-style self-review instructions, test-runs it, and hands over a working automation. ## Quick Start Help me automate my weekly reporting workflow by designing and building an agentic AI system for it.
