start-idealisted

# Start a new learning idea Take the user from "no idea" (or a rough one) to a committed set of specs a milestone build can execute from. This is the front half of `PROCESS.md`; the back half is the `build-milestone` skill. Follow a brainstorming HARD-GATE: **write no implementation code until specs are approved.** ## When this triggers "Let's brainstorm a new idea", "I want to build something to learn <concept>", "start a new project", or picking up a fresh idea in the workspace. ## Procedure ### 1. Ground the learning goal - Confirm which concepts the user wants to internalize. - Check `concepts/` for what's already mapped. - Anchor to a real, existing codebase the user can study where possible. ### 2. Brainstorm product ideas (one question at a time) - Propose 3–4 *product* ideas that NATURALLY exercise the target concepts — new products, not clones of the anchor. For each: the pitch, the "wow" moment, and which concept it showcases. - Let the user pick or riff. Prefer multiple-choice questions; one at a time. ### 3. Sharpen the chosen idea - Write a north-star scenario (the happy path). Identify where the hard concept earns its keep vs. a trivial version. Set a deliberately tight v1 scope (YAGNI). - **Gate:** get explicit approval of the vision before writing the brief. ### 4. Commit the product brief - Create `projects/<name>/README.md`: vision, north-star scenario, why-it-needs-the-concept, v1 scope (in/out), locked decisions table, milestone map. - Add a