agentsclimarketplace

Site preflight

Skill tjrice42/design-judgment-skills/site-preflight

Ten Claude skills that encode a design practice: journaling, critique, voice, and a ship verdict. Judgment, installed.

Install
npx -y skills add tjrice42/design-judgment-skills --skill site-preflight

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

2 things to look at

  • no licenseNo license file was found in the repository. Code published without one is not open source by default, so using it at work is a question for whoever answers licensing questions where you are.
  • 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

Run a mechanical pre-flight QA check on a portfolio site or any live website before sharing it — broken links, mobile rendering, load weight, meta/OG tags and link-unfurl preview, favicon, placeholder text, console errors, and accessibility basics. Use this skill whenever the user is about to share, submit, or publish a site, asks "is my site ready," mentions applying with a portfolio link, or wants a link check or site QA. Cheap and fast — when in doubt, run it.

SKILL.md

3.9 KB, as published. Nobody here has run it

Site Preflight

A mechanical, pass/fail QA sweep of a live site before anyone important sees it. This skill checks function, not content — it has no opinions about design quality or storytelling (that's portfolio-critique / portfolio-edit-pass). Its job is making sure nothing embarrassing and avoidable happens: the broken deep link, the blank OG preview in a recruiter's Slack, the lorem ipsum in the footer.

Workflow

1. Crawl

Run scripts/preflight.py <url> (Python 3, stdlib only — no installs needed). It crawls every same-domain page reachable from the start URL, checks every internal and external link, and extracts meta/OG data and asset weights per page. If the script can't run in the current environment, replicate its checks manually with fetches.

2. Checks — every item gets ✅ PASS / ❌ FAIL / ⚠️ WARN

Links & navigation

  • Every internal link resolves (no 404s, no redirect loops). Deep links into case-study sections included.
  • Every external link resolves. WARN on redirects to unexpected domains.
  • Nav is consistent across pages; no orphan pages unreachable from nav.

Link-share preview (how the URL unfurls in Slack/iMessage/LinkedIn when a recruiter shares it)

  • og:title, og:description, og:image present on every page, and the values actually describe the person/work — FAIL on framework defaults ("Vite + React"), empty strings, or localhost image URLs.
  • <title> and meta description set per page; favicon present and loading.

Content hygiene

  • No placeholder text anywhere: lorem ipsum, "TODO", "coming soon", "[name]", template boilerplate. Scan rendered text, not just source.
  • No broken images (404s, zero-byte, alt-less hero images).
  • Copyright year / "last updated" not stale.

Weight & speed

  • Total page weight per page; FAIL any page > 5 MB, WARN > 2 MB. Flag individual images > 500 KB with the specific file (uncompressed PNG exports are the usual culprit).
  • WARN on render-blocking patterns visible in the HTML (giant inline payloads, unoptimized font loading).

Mobile

  • Viewport meta tag present.
  • Fetch as mobile UA; flag fixed-width layouts, horizontal overflow signals, and tap targets that are clearly desktop-only (hover-dependent nav).

Accessibility basics (not a full audit — the embarrassment-prevention tier)

  • Images have alt text; page has one h1; heading levels don't skip wildly; links have discernible text (no bare "click here" or icon-only links without labels); html lang attribute set; obvious contrast disasters flagged.

Console & errors

  • Any 4xx/5xx on page assets (JS bundles, CSS, fonts) — these usually mean something visibly broken.

3. Report

Deliver preflight-report.md:

  1. Verdict line first: "READY TO SHARE" or "N blockers — fix before sharing" (blockers = any FAIL).
  2. FAILs with the exact URL/file/line evidence and the fix, each sized (5-min / 30-min).
  3. WARNs grouped after, clearly optional.
  4. The unfurl preview: render what the link will look like when pasted in Slack (title / description / image) so the user sees what a recruiter sees.

Rules

  • Evidence for every flag: exact URL, exact asset, exact text found. No "some links may be broken."
  • No design opinions. If tempted to comment on aesthetics or content quality, don't — route the user to the critique skills instead.
  • This check is cheap; recommend re-running after any deploy and always immediately before submitting an application.

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.