Menu
Deploy Butler to production (Cloudflare Workers API + Pages frontend) with pre-flight checks and a post-deploy /health smoke check
Install with Codex or Claude Copy this prompt, paste it into Codex, Claude, or another assistant, and let it review the skill page and install it for you.
Based on SOC occupation classification
| name | deploy |
| description | Deploy Butler to production (Cloudflare Workers API + Pages frontend) with pre-flight checks and a post-deploy /health smoke check |
| disable-model-invocation | true |
Ship Butler to production: deploy the @butler/api worker and the @butler/frontend
site to Cloudflare, then confirm the API is live with a /health smoke check.
Human-invoked only. Deploying is a side-effecting, production-facing action. This skill has
disable-model-invocation: trueand must NEVER run automatically — only when a human explicitly runs/deploy. Onmain, CI already auto-deploys on push (.github/workflows/ci.yml); use this skill for an intentional manual deploy.
CLOUDFLARE_API_TOKEN and CLOUDFLARE_ACCOUNT_ID set in the environment (same secrets
CI uses). If unset, stop and ask the human to provide them — do not read them from any
service .env.Refuse to deploy unless all of these hold (report which failed and stop):
git status --porcelain is empty (no uncommitted changes).main and up to date — git rev-parse --abbrev-ref HEAD is main, and
git fetch origin && git status -sb shows no divergence from origin/main.gh run list --branch main --limit 1 --json conclusion,headSha reports success
for the current HEAD SHA. Never deploy a commit whose CI failed or is pending.pnpm --filter @butler/shared build && pnpm run type-check && pnpm run lint.cd packages/api && pnpm run deploy # = wrangler deploy (worker: butler-api)
Capture the deployed URL that wrangler prints (e.g. https://butler-api.<subdomain>.workers.dev).
Save it as API_URL for the smoke check. If you cannot parse it from the output, run
pnpm exec wrangler deployments list or ask the human for the production API URL — do
not guess a URL.
cd packages/frontend && pnpm run build
pnpm exec wrangler pages deploy dist --project-name=butler-frontend
Note the Pages URL wrangler prints for reference.
Hit the live API health endpoint (GET /health returns { "status": "ok" }):
curl -fsS "$API_URL/health"
{"status":"ok"}.status is not "ok".Report the outcome explicitly, e.g.:
✅ Deploy OK — API healthy at <API_URL>/health (status: ok); frontend at <PAGES_URL>❌ Smoke check FAILED — <API_URL>/health returned <code>/<body>. Investigate before announcing the deploy; consider rolling back (Cloudflare dashboard or wrangler rollback).deploy job in
.github/workflows/ci.yml — keep them in sync if CI changes.main.Scaffold a new golden case for @butler/evals — validate and append the case to golden.jsonl, re-capture the baseline, and confirm the eval gate still passes. Use when a real agent failure or escaped bug should become a regression benchmark.
Work the Butler ready-for-agent issue queue in priority order — select the highest-priority unblocked issue (P0 > P1 > P2) and run /implement-issue on it, looping up to a per-run cap. Use ONLY when explicitly asked to "work the queue" / "implement the next issue(s)", or when fired by the scheduled routine. Do not auto-run.
Break a plan, spec, or PRD into independently-grabbable issues on the project issue tracker using tracer-bullet vertical slices. Use when user wants to convert a plan into issues, create implementation tickets, or break down work into issues.
Turn the current conversation context into a PRD and publish it to the project issue tracker. Use when user wants to create a PRD from the current context.
End-to-end implementation of a GitHub issue — from reading the issue through TDD, code review, PR creation, CI monitoring, and merge. Use when the user says "implement issue
Phase 9 retrospective — read a window or set of recently-merged PRs, classify each harness/run finding with the eval-failure taxonomy, and append tagged learnings to docs/retro/learnings.md. Propose-only — writes ONLY under docs/retro/. Use when explicitly asked to "run a retrospective", "retro the last N PRs", or to review a batch of merged work for harness lessons. Do not auto-run.