agentleFS
nhattrung0911nhattrung0911/shipwright/skills/shipping-production-websites/SKILL.md

shipping-production-websites

Skill0 starsChanged 5 months agoMITView on GitHub

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.

The file

We split this skill into the rules it gives an agent

4 rules. About 0 are common, 4 are distinctive, and 0 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈4

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 0

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Run in order. Each step hands to a superpowers skill.

    line 18
  2. Distinctive:

    Every shipped site needs a decision on each.

    line 73
  3. Distinctive:

    Do NOT call a site production-ready until:

    line 94
  4. Distinctive:

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

    line 103

≈ “Common” is an estimate: at least 1 in 5 of the 372,429 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

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. Requirements → REQUIRED 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/UX → REQUIRED 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. Plan → REQUIRED 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. Build → REQUIRED 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

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.