← ClaudeAtlas

devops-opslisted

Monitors the health of the live site and CI/CD pipeline — checks deployment status, reads build and error logs, confirms all environment variables are correctly set. Also includes the step-by-step procedure to roll back a broken production deployment in under 60 seconds without writing any code. Use when the operator says "is the site up", "deployment failed", "check the logs", "roll back", or anything about production health.
talentedgeai/infinite-leverage · ★ 0 · DevOps & Infrastructure · score 62
Install: claude install-skill talentedgeai/infinite-leverage
# DevOps: Operations ## Scope **In scope:** - GitHub CI/CD: Actions workflows, branch protection, PR checks - Vercel: deployment status, build logs, runtime logs, environment variables - Production health monitoring via vercel CLI **Out of scope:** - Writing or reviewing application code (Developer owns this) - Content pipeline (Writer/Designer/Web Publisher own this) - Database schema changes (escalate to human engineer) ## Vercel CLI — Read-Only Monitoring ```bash vercel ls # list recent deployments + status vercel inspect <deployment-url> # deployment details + build info vercel logs <deployment-url> # runtime logs vercel env ls production # confirm all env vars present ``` ## Vercel CLI — Management (require explicit user confirmation) ```bash vercel env add KEY production # add environment variable vercel link --project {slug} # link local dir to Vercel project ``` Never run `vercel deploy` or `vercel --prod`. All deployments through `git push` → CI/CD only. ## Deployment Model - All deployments flow through GitHub → Vercel CI/CD only - Never run `vercel deploy` or `vercel --prod` directly - Never push to `main` — all changes through PRs - Vercel CLI for read-only operations only; writes through CI/CD ## Production Rollback (when a deployment breaks the live site) This is the fastest fix. Tell the operator these exact steps: 1. Go to **vercel.com** → open the project → cli