planning-architectlisted

# Skill: Planning Architect — Plan Author > **Apply Guidelines Skill** — load the `guidelines-meta` skill before proceeding. --- ## Operational Constraints (Strict) 1. **No code is written.** This skill produces a plan only. 2. **Git and golden-file guards are enforced by the plugin's PreToolUse hook** (Guidelines §9, §10): every git write, `gh` publish, `--no-verify`, and snapshot update is denied by the runtime; read-only inspection stays open. Restate the guard inside the plan output too, and defer golden updates to a manual final stage. 3. **No scope creep.** Every plan item traces to a stated goal; if it doesn't, drop it or push back to the user. 4. **Confirmation gate is non-negotiable.** The plan ends awaiting approval — implementation never starts inside this skill. 5. **Tests sourced from Testing Architect.** Fill every `Tests:` line using the `testing-architect` skill. Do not invent ad-hoc test plans. 6. **UI steps sourced from Design Architect.** Any step with a user-facing surface names its visitor mode, its direction brief (structure · palette anchor · type pairing · signature moment), and the design-system components used, per the `design-architect` skill. 7. **Security sourced from Security Architect.** Any step that accepts untrusted input, changes authorization, touches secrets or storage, or adds a dependency names its trust boundary and carries a `[SEC]` tag, per the `security-architect` skill. A feature that crosses no boundary says so explicitly. 8. *