ship-check
Skill141 starsChanged 34 days agoUnlicenseView on GitHub
Pre-launch gate for this starter — build, mechanical rule checks, SEO, metadata and brand-kit completeness, the robot form, measured load (Lighthouse people + robot) and scroll (PC + phone), real-phone behaviour, accessibility, env and secret hygiene, the deploy steps and a closing retro. Use when the user says "ready to ship", "deploy this", "launch checklist", "is this production ready", or before handing a site to a client.
- Reads credentials
19 other projects use a version of this.
Adopted by
Most recently taken up by
- Archive228/claude-fable-rea-blatheir own version, first seen Nov 15, 2026
- nadavnaveh/skillstheir own version, first seen Nov 8, 2026
- kitloomlk-stack/nextjs-supabase-kittheir own version, first seen Oct 15, 2026
- joetawil7/first-passtheir own version, first seen Oct 14, 2026
- fightingforsidewalk/vault-relaytheir own version, first seen Oct 13, 2026
- AhmedMerza/agent-skillstheir own version, first seen Oct 11, 2026
Used with
Projects with it also use review.
Other files people use alongside this one.
Files found in the same public projects as a version of this one, more often than they turn up anywhere else. Each project counts once, large collections count for less, and this file's own repository is left out.
SentrySkillreview
Fixture review skill.
Also in 3 projects that use this
Other versions19 projects use a version of this.
Word for word · 0
None found.
Edited · 0
None found.
Wrote their own version · 19
- facontidavide/PlotJuggler★ 6.3ksame name
- joetawil7/first-pass★ 128same name
- Archive228/claude-fable-rea-bla★ 18same name
- mohankrishnaalavala/context-router★ 11same name
- Aximande/agent-skills★ 4same name
- koach08/claude-code-skills★ 4same name
- nadavnaveh/skills★ 3same name
- AhmedMerza/agent-skills★ 2same name
- AhmedMerza/claude-skills★ 2same name
- fightingforsidewalk/vault-relay★ 2same name
9 more not listed
How matches are found. Word for word: the exact same text. Edited: shares passages with this file (“99% same” means 99% of its five-word phrases also appear here; 10% or more counts). Their own version: similar meaning, or, weakest, the same name. We can’t tell which project had it first. Projects are counted among the 200 most-starred candidates, and mentions are links to the repository we found, up to 50 per site, so both counts are a floor.
In this repository20 other files it gives its agents.
- AGENTS.mdAGENTS.md
- CLAUDE.mdCLAUDE.md
- aeo-visibilitySkill
- figma-to-sectionSkill
- i18n-seoSkill
- local-seoSkill
- mobile-device-qaSkill
- optimize-3d-sceneSkill
- optimize-loadSkill
- optimize-performanceSkill
- payload-adminSkill
- payload-analyticsSkill
- payload-cmsSkill
- qa-verifySkill
- schema-markupSkill
- seo-auditSkill
- seo-growSkill
- site-migrationSkill
- supabase-authSkill
- supabase-dbSkill
The file
We split this skill into the rules it gives an agent
10 rules. 0 are common, 2 are distinctive, and 8 are very specific to this project.
0
Common
Most of the 18 edited versions we read keep these.
2
Distinctive
This author's own choices. Worth a look.
8
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 9
Run in order. Anything unchecked is stated as unchecked in the summary — never assume a pass.
- Very specific: line 22
Run the qa-verify skill across the site — every route, every breakpoint.
- Very specific: line 27
Run the seo-audit skill. Non-negotiable before launch:
- Very specific: line 29
every public route is in sitemap.ts.
- Very specific: line 50
if this replaces an existing site → the site-migration skill's redirect map is live and verified
- Very specific: line 78
Run the mobile-device-qa skill's close-out: iOS toolbar probe and context-loss probe for any live scene, scene fps (no fixed phone caps), …
- Very specific: line 100
Every secret is server-only via getServerEnv().
- Distinctive: line 102
Every production env var is actually set on the host — a missing one fails the zod parse at boot, which is the intended behaviour, so check …
- Very specific: line 112
if Payload is installed: payload migrate has run against the production database, and push is false in production
- Very specific: line 124
Update obsidian/meta/changelog.md with the launch.
“Common” is measured: kept by at least 6 in 10 of the 18 edited versions of this file we could read. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
Ship check
Run in order. Anything unchecked is stated as unchecked in the summary — never assume a pass.
1. It builds and it obeys the rules
yarn lint
yarn build # must pass, no warnings you cannot explain
.claude/scripts/verify.sh # zero FAILs2. It is correct
Run the qa-verify skill across the site — every route, every breakpoint.
Then click through the real thing: yarn build && yarn start, not just dev.
3. It can be found
Run the seo-audit skill. Non-negotiable before launch:
- every public route is in
sitemap.ts; everyhrefresolves (no 404s from nav items or the banner's/privacy-policy) robots.tsallows crawling and no stagingnoindexsurvivesNEXT_PUBLIC_SITE_URLset in the production environment — and on the deployed site the canonical names the real domain andcurl -sI <og:image>returns 200- no
TODO:placeholder insrc/lib/site.ts(a production build fails on them by design); the brand kit is the site's own — favicon set from its logo, a 1200×630 share image you looked at (tools/qa/brand-kit.mjs) - unique title + description per route
- the robot form:
curl -A "Googlebot/2.1"and-A "GPTBot/1.0"get<meta name="x-robot-view">, the same title/canonical/<h1>, no text under ahiddenancestor; the human/is still prerendered (optimize-load/references/robot-path.md) Organization+WebSiteJSON-LD renders and validatesnode tools/qa/seo-crawl.mjs --url https://<deploy>/→ zero FAILs: every sitemap page 200, unique filled titles, real canonicals, one h1, JSON-LD parses, the robot form matches the people form;/llms.txtlists the sitemap's pagessiteConfig.aiCrawlersis the client's stated policy (not the default by accident); Search Console + Bing verified (GOOGLE_SITE_VERIFICATION,BING_SITE_VERIFICATION), the sitemap submitted- if this replaces an existing site → the
site-migrationskill's redirect map is live and verified
4. It is fast
Measure, do not guess — against the deployed URL (localhost is a floor:
one site's mobile LCP was 4.8 s local, 31 s hosted). If a WebGL/three.js scene
exists → the optimize-3d-scene skill first.
node tools/qa/lighthouse.mjs --url https://<deploy>/ --runs 3 # people, mobile + desktop
node tools/qa/lighthouse.mjs --url https://<deploy>/ --as-bot # the robot form (what PageSpeed sees)
node tools/qa/scroll-test.mjs --url https://<deploy>/ --first-scroll # PC wheel + phone touch, cold + warm, + the unlock moment
node tools/qa/axe-sweep.mjs --url https://<deploy>/ # contrast through the entrance
node tools/qa/check-motion.mjs --url https://<deploy>/ # reduced motion + robot don't hang
node tools/qa/seo-crawl.mjs --url https://<deploy>/ # every page as a crawler reads it — zero FAILs- Performance ≥ 90 on mobile and desktop; Accessibility, Best Practices, SEO
= 100 — people and robot (
optimize-load). A gap is stated with its number and cause. - Scroll
idealon PC and phone: no frame > 50 ms, ≤ 1 % dropped (optimize-performance). Lighthouse never scrolls. - LCP ≤ 2.5 s, CLS ≤ 0.1, INP ≤ 200 ms;
grep -l ZodError .next/static/chunks/*.jsprints nothing.
5. It works on a phone in a hand
Run the mobile-device-qa skill's close-out: iOS toolbar probe and
context-loss probe for any live scene, scene fps (no fixed phone caps), the menu
in light and dark scheme, overflow at 320–430, touch sliders, WebKit for
anything iOS-only. Then push a preview and have the client (or a teammate) open it on their
phone, naming what to try. State what could only be checked in emulation.
6. It is usable by everyone
- Keyboard-only pass through every interactive element; focus always visible
prefers-reduced-motionhonoured — content readable with motion off- Contrast AA; alt text everywhere; landmarks named
- 44px touch targets; no horizontal scroll at 320px
tools/qa/resize-check.mjs— live resizes and device-preset switches land where a fresh load does
7. Nothing leaks
grep -rniE "sb_secret|service_role|BEGIN (RSA|PRIVATE)" src/ .env.example
git log --oneline -20 # no secrets in history- Every secret is server-only via
getServerEnv(); nothing sensitive isNEXT_PUBLIC_ .envis gitignored;.env.exampledocuments every key with placeholder values- Every production env var is actually set on the host — a missing one fails the zod parse at boot, which is the intended behaviour, so check before launch not after
8. Deploy
Vercel is the default target for this stack (vercel → vercel --prod, or a Git
integration). Confirm after deploying:
- production env vars set for the Production environment, not just Preview
- custom domain + HTTPS resolving,
www/apex normalised one way - if Payload is installed:
payload migratehas run against the production database, andpushis false in production - a real page loads, a form submits, an image from storage renders
- a Git-connected deploy that stops updating production: check the commit
author's email belongs to the Vercel team (a repo-local
user.emailoutside the team blocked every deploy on three sites for months)
9. Hand over — and retro
Summarise: what was checked and passed, what failed and was fixed, what could not
be verified and why, plus the measured numbers (the grid, the scroll verdicts).
Update obsidian/meta/changelog.md with the launch.
Then a short retro, for the next project cut from this starter: a measured fix
worth reusing → obsidian/knowledge/fix-catalog.md (symptom, fix, before →
after, rule/observed); something that misled you → obsidian/knowledge/pitfalls.md;
a defect that came from the starter itself → note it there as a starter fix
candidate; a skill that was wrong or silent → edit it.
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
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.

