What problem does it solve? When facing an open problem — a product decision, an onboarding flow, a naming question, or a technical direction — teams often jump on the first idea instead of the best one. This Skill runs a disciplined brainstorming session that deliberately generates many options, clusters them into real directions, stress-tests each, and records why every rejected idea lost. ## Core Features & Use Cases - Structured divergence: Generates at least 8 candidates across at least 3 named thinking lenses (SCAMPER, inversion, Six Thinking Hats, analogy, constraint-flip, first-principles, repo precedent) with critique switched off during generation. - Converge without deciding for you: Clusters candidates into 3-5 directions, pre-mortems each one, states a recommendation, then waits — the user always makes the final pick, and every menu includes an open slot for the user's own idea. - Persistent session document: Writes a brainstorm-session.md capturing the problem, all candidates, directions, the pick, and the reasons every loser lost, so future sessions never re-propose dead ideas. - Use Case: Ask "how should we onboard new merchants?" and receive a framed problem statement, a dozen lens-generated candidates, four stress-tested directions, and a written record of the chosen direction with tagged constraints ready to hand off to analysis or planning lanes. ## Quick Start Ask the assistant to run /orc-brainstorm on a problem like "our support queue is drowning us" and let it generate and stress-test options before you pick one.
