What problem does it solve? When a client's live site, bot, or service goes down and you are handling it alone — often from a phone — generic incident-response playbooks waste the critical first minutes on org-style role assignments and status meetings. This Skill gives a single operator a concrete diagnostic order, ready-to-send client message templates in plain Hebrew, and a one-paragraph postmortem format. ## Core Features & Use Cases - Prioritized diagnostic sequence: Check what changed in the last hour first, then Render, Cloudflare Pages, MongoDB Atlas, the Telegram webhook, and Sentry — ordered by probability, with platform-specific traps like stale last_error_message values and free-tier cold starts. - Client communication templates: Three pre-written messages (first response within 10 minutes, status update, closure) that promise an update time rather than a fix time, in jargon-free language. - Severity triage and rollback safety: Three severity levels instead of SEV1-4, plus a data-touch check before rolling back a deploy so a migration is not reverted blindly. - Use Case: A client messages that their site is down at 2 AM. The Skill tells you to send the first holding message immediately, then walk the diagnostic ladder — recent changes, Render status, Cloudflare deploy, Atlas connections — and close with a one-paragraph postmortem. ## Quick Start Use the incident-response skill to help me handle a production outage my client just reported on their live site.
