agentsclimarketplace

Shipping production websites

Skill nhattrung0911/shipwright/skills/shipping-production-websites

Production-grade engineering discipline for AI coding agents — 5 composable skills (plan, build, secure, operate) for Claude Code, Codex & Gemini. Never skips a step, never fakes done.

Install
npx -y skills add nhattrung0911/shipwright --skill shipping-production-websites

Assembled from the repository path, not quoted from the project. Check it against their README if it does not work.

One thing to look at

  • 0 stars0 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.

What its author says it does

Copied from the file, not written here

Use when starting, planning, scoping, or auditing any website or web app intended to reach real users — greenfield builds, "build me a site/dashboard/SaaS/landing page", adding a major feature, or pre-launch production-readiness checks. Stack-agnostic.

SKILL.md

8.8 KB, as published. Nobody here has run it

Shipping Production Websites

Overview

Single entry-point gate + checklist for building websites/web apps to a production standard. This skill does not replace superpowers — it sequences them and adds the production-readiness checklist that generic planning misses.

Core principle: A site is "done" only when every one of the 14 production pillars below has an explicit decision (even if the decision is "deliberately skipped, here's why"). Silence on a pillar = a gap, not a default.

Announce at start: "Using shipping-production-websites to gate and plan this build."

The Flow (coordinate with superpowers)

Run in order. Each step hands to a superpowers skill; this skill supplies the web-specific content.

  1. RequirementsREQUIRED SUB-SKILL: superpowers:brainstorming. Lock: who uses it, scale, must-have vs later, hard constraints (budget, compliance, deadline). Don't pick stack silently — surface options.
  2. Architecture decisions → fill the Architecture table below. One line per choice + why.
  3. Production checklist → walk the 14 pillars. Mark each: ✅ in-scope / ⏭️ deferred (+reason) / ❌ N/A (+reason).
  4. UI/UXREQUIRED SUB-SKILL: ui-ux-pro-max for layout, design system, components, a11y-aware states. The interface is the product for a website — do not defer this to a footnote.
  5. PlanREQUIRED SUB-SKILL: superpowers:writing-plans. Embed pillar decisions as acceptance criteria, not afterthoughts. Big build = split per subsystem.
  6. Isolate (if non-trivial) → REQUIRED SUB-SKILL: superpowers:using-git-worktrees.
  7. BuildREQUIRED SUB-SKILL: superpowers:test-driven-development. Red-green-refactor per task. Keep context lean throughout: REQUIRED SUB-SKILL: token-frugal-engineering (delegate tests/logs/search to subagents).
  8. Verify before "done"REQUIRED SUB-SKILL: superpowers:verification-before-completion, then run the /code-review and /security-review slash commands. Evidence, not assertion.

Skip a step only when its superpowers skill itself says it doesn't apply.

If disciplined-delivery is already active, this skill fills its web row — don't re-run Frame/Plan; supply the web-specific pillars below into its loop.

Portability: the superpowers:*, ui-ux-pro-max, and /code-review-style references are Claude Code sub-skills/commands. On Codex/Gemini they may not exist — treat each as "(if installed)"; the pillars and gates below stand alone and are the portable core.

Architecture Decisions (fill before planning)

DecisionPick + why
RenderingSSR / SSG / SPA / hybrid — why
Framework / lang
Data storeSQL / NoSQL / serverless — why
Auth modelsessions / JWT / OAuth / managed (Clerk/Auth0)
Hosting / deploy targetVPS / PaaS / serverless / edge
State of multi-tenancysingle / multi-tenant
Sync vs async workjobs/queues needed?

Full-Stack Layer Map (what "full-stack" actually includes)

A site is not just frontend + backend. Decide each layer or mark N/A — these are the buildable parts:

LayerCoversCommon miss
Config & env12-factor: per-env config (dev/stage/prod) separate from secrets; no config baked into build; feature flagsconfig-vs-secret confusion; prod config in repo
i18n / l10n (decide or N/A)locale routing, translations, RTL, timezone/currency formattinghardcoded strings; breaks for non-US users
Payments (if any)webhook signature verify + idempotency; never store raw card (PCI scope); refund/dispute pathdouble-charge on retry; storing PAN
FrontendUI components, routing, client state, forms, loading/error/empty stateserror & empty states forgotten
Backendbusiness logic, services, domain rules, validationlogic leaking into controllers
APIendpoints, contract, versioning, auth, rate-limit (pillar 3)no input schema
Databaseschema, indexes, migrations, backups (pillar 2)no indexes, irreversible migrations
Authlogin + per-resource authorization (pillar 4)authn without authz
File/mediaupload, validation, storage (S3-class), CDN deliveryunvalidated uploads
Background workqueues, scheduled jobs, retries, idempotencysync work that should be async
Realtime (if needed)websockets/SSE, presence, reconnectionN/A for many sites
Email/notificationstransactional email, deliverability (pillar 12)SPF/DKIM unset → spam folder
Cachingclient/CDN/server/DB cache + invalidation (pillar 10)cache with no invalidation plan
Infra/DevOpshosting, CI/CD, env config, secrets (pillars 8, 14)manual deploy, secrets in code
Observabilitylogs, error tracking, uptime (pillar 9)bolted on after an incident

Each layer maps to the pillars below for its "done" bar. Skip a layer only with a written reason.

The 14 Production Pillars

Every shipped site needs a decision on each. Don't confuse "I didn't think about it" with "not needed."

#PillarMinimum bar to clear
1ArchitectureRendering + data flow + boundaries decided and written (ADR-style)
2Data & migrationsSchema versioned; migrations reversible; no manual prod edits; backups
3API contractVersioned, validated input (schema), consistent errors, pagination, rate-limit
4AuthN / AuthZLogin + per-resource authorization (not just authentication); least privilege
5SecurityOWASP Top 10:2025 — broken access control, injection, misconfig, supply chain, secrets in env not code, HTTPS/headers. Depth: REQUIRED SUB-SKILL securing-applications
6Validation & errorsServer-side validation always; graceful error states; no stack traces to users
7TestingUnit + integration + e2e covering EACH critical journey (incl. auth + payment), not just one happy path; CI runs them on every PR
8CI/CDAutomated build→test→deploy; rollback path; no manual ftp
9Observability & opsStructured logs + error tracking + uptime/health check + alerting to a person. Runtime controls (rate limit, timeouts, retries, graceful shutdown) + Day-2 maintenance depth: REQUIRED SUB-SKILL operating-production-services
10PerformanceCore Web Vitals budget (default LCP<2.5s, INP<200ms, CLS<0.1); caching strategy; image/asset optimization; DB indexes
11Accessibility, SEO & analyticsWCAG AA basics (contrast, labels, keyboard); semantic HTML; meta/OG, canonical URLs, robots, sitemap, structured data, redirect map; consent-gated analytics
12Domain & deliveryDNS configured; TLS cert issuance + auto-renewal; transactional email deliverability (SPF/DKIM/DMARC) if app sends mail
13Cost & quotasCloud budget alarms; third-party API quota/rate limits understood; no surprise-bill failure mode
14Compliance & opsData-privacy: PII inventory, DSAR (deletion/export), cookie/consent gate, retention (depth in securing-applications); legal pages; secrets mgmt; runbook

Pre-Launch Gate

Do NOT call a site production-ready until:

  • Every pillar marked ✅ / ⏭️(+reason) / ❌(+reason) — no blanks
  • Tests green in CI; /security-review clean or risks accepted in writing
  • Rollback tested; backups verified restorable
  • Error tracking + uptime check live
  • Secrets out of source; prod env separate from dev

Escalation: installable vendor skills

When a pillar needs depth beyond this checklist, pull the specialist skill instead of improvising:

  • DevOps/IaC/observability (Docker, K8s, Terraform, CI, Prometheus): /plugin marketplace add akin-ozer/cc-devops-skills
  • Full-stack practical (tests, db migrations, API scaffold, a11y, git): /plugin marketplace add jezweb/claude-skills
  • Official vendor skills (Stripe payments, Neon Postgres, Cloudflare perf, Trail of Bits SAST, Playwright): see officialskills.sh / VoltAgent/awesome-agent-skills
  • UI/UX: ui-ux-pro-max (already installed)

Common Mistakes

MistakeFix
Pick framework before requirementsBrainstorm first (step 1)
AuthN without AuthZPillar 4 — authorize every resource access
Tests added "after it works"TDD (step 6); tests-after prove nothing
"Works on my machine" → shipCI gate (pillar 7/8) before deploy
Observability bolted on post-incidentPillar 9 in plan, not after fire
Treating a skipped pillar as a defaultEvery pillar gets an explicit written decision

Gives 0 of the 12 instructions most ship operate skills give

Counted across 779 of the 1,178 authors here whose files we hold, read 2026-08-06

  • document a rollback plan before deploymentin 40 of 779, across 21 files
  • create an annotated git tagin 21 of 779, across 20 files
  • Run the test suitein 20 of 779
  • update the changelogin 20 of 779, across 18 files
  • verify deployment health after launchin 19 of 779, across 10 files
  • clean up feature flags after full rolloutin 18 of 779, across 10 files
  • verify the working tree is cleanin 18 of 779
  • test both feature flag statesin 17 of 779, across 9 files
  • Make database migrations backward-compatiblein 16 of 779, across 8 files
  • set up error monitoring before launchin 15 of 779, across 7 files
  • monitor metrics at each rollout stagein 14 of 779, across 5 files
  • create a github releasein 14 of 779

Said here and by no other author read

  • brainstorm requirements before choosing a stack
  • fill the architecture decisions table
  • evaluate all 14 production pillars
  • delegate ui/ux work to a sub-skill
  • write a plan embedding pillar decisions as acceptance criteria
  • use test-driven development for the build

Grouped from the skills themselves: near-identical wordings counted once, and counted by distinct author, so one author publishing three of these counts once.

Keep looking

Skills are one crate of 328,083. Ordering is by how many stacks a row turns up in, so the top of any crate is what has actually been picked rather than what has the most stars.