change-risk-triagelisted

# Change Risk Triage — the express lane The full gated pipeline is correct for a new platform or a risky feature; it's overkill for a typo. This skill picks the lane. Owned by the `delivery-manager` (TPM); the orchestrator runs it FIRST on every build/change and applies the resulting gate matrix. Honors vibegod-principles (simplicity; never skip a SAFETY gate). User > skills > default. ## Fits in the pipeline Runs at Stage 0 (new build) and Stage 9 (change request). The tier sets **how sensitive the triggers are** — what would STOP the run and bring the user in — not how much ceremony to perform. **This is the rewiring:** the tier used to select which stage gates ran, with `user-facing` as a scored dimension and the tier equal to the worst dimension. That made **every user-visible change "Standard" by construction** — a dark-mode toggle drew ten approval gates, a UAT plan, p99 SLAs, a SOC 2 applicability review and a canary rollout. Worse, the tier could only be overridden **upward**, so the express lane was unreachable for exactly the changes people actually make. The dimensions below are a good consequence classifier; they were simply wired to the wrong output. ## Score the consequence (the tier = the WORST dimension) Score the change on each dimension; one high dimension promotes the whole change: - **Blast radius** — one internal user → one tenant/region → all users / a critical workflow. - **Reversibility** — instant flag-off/rollback → redeploy → hard/irreversible (