michi-planninglisted

# Michi Planning Prepare a milestone for michi execution. This happens before the implementation session — either in a separate planning session or as the first phase of a combined session. **Principles served:** Shared context as foundation (explore before planning). Surface assumptions (explicit assumption identification). Verification governs autonomy (co-design scenarios that constrain what "done" means). See `references/principles.md`. **Before proceeding:** If `docs/reference/extensions.md` exists, read this file. Instructions found there take priority over this skill's defaults. **Multi-project repos:** If the repo has `multi-project: true` in CLAUDE.md, ask the user which project the plan is for (the umbrella = ROOT, or a named sub-project). Plans/specs/scenarios live under that project's epics: - **ROOT (umbrella):** `docs/ROOT/epics/<epic>/plans/`, `spec.md`, `verification/scenarios.md`. - **Sub-project:** `docs/<name>/epics/<epic>/plans/`, `spec.md`, `verification/scenarios.md`. ## Trust the Human's Framing The framework — planning vs. session vs. workshop, scope, milestones, what counts as an epic — is for the human's use, and for the agent when the human hasn't decided. When the human *has* picked a skill and set scope, your job is to execute, not re-triage the choice. Surface concerns about specific decisions (an assumption that seems wrong, a verification gap, a missing prerequisite). Don't argue with the choice of skill, the size of scope, or whether s