agentsclimarketplace

World class web release

Skill frankxai/starlight-design-intelligence/skills/world-class-web-release

Use for any high-value website, landing page, homepage, partner page, product surface, brand experience, or redesign where language, typography, imagery, motion, implementation, and production proof must feel distinctive and human. Also use when upgrading a web-design quality system or release gate.From its SKILL.md

Install
npx -y skills add frankxai/starlight-design-intelligence --skill world-class-web-release

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.
  • 1 stars1 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.

SKILL.md

4.8 KB, 938 tokens by cl100k_base, as published. Nobody here has run it

World-Class Web Release

World-class is a release property, not a styling adjective. A surface passes only when its language, composition, type, imagery, motion, engineering, and production evidence tell one coherent story.

Required Sequence

  1. Reconstruct context. Read the repository contract, brand pack, existing surface, source material, prior decisions, and known constraints. Name the recipient, their knowledge state, the job of the surface, and the consequence.
  2. Capture current truth. For a redesign, capture the existing desktop and mobile states. For a greenfield route, capture the host site's desktop and mobile context. Gather real workflows, artifacts, quotations, results, and authorized imagery. Stop when a referenced existing surface cannot be captured.
  3. Write the experience thesis. State one recipient, job, promise, primary action, signature proof, and intended feeling.
  4. Compare exactly three directions. Each must materially differ in composition, typography, imagery, interaction, and motion posture. Select one before implementation.
  5. Build the static composition. Resolve hierarchy, pacing, typography, proof, responsive behavior, and conversion before motion.
  6. Pass editorial articulation. Use skills/editorial-articulation/SKILL.md. Read every sentence aloud. Reject name-swappable or taxonomy-led copy.
  7. Earn motion. Use skills/motion-and-interaction/SKILL.md. Motion must orient, explain, confirm, reveal causality, or create one brand memory.
  8. Compare and verify. Capture desktop, mobile, interactions, and reduced motion. Test accessibility, reflow, loading, performance, links, claims, privacy, analytics, and failure states as applicable.
  9. Separate responsibility. Maker, verifier, and approver are distinct for public flagship releases.
  10. Prove production. Record commit, checks, preview, production URL, post-deploy verification, the paths covered by the evidence, and exact rollback in the release manifest. Generate this receipt after deployment. Keep the manifest outside the production commit—such as a CI artifact or a later receipts commit—because a commit cannot contain its own SHA. Store blocking evidence beside the manifest and content-address every artifact with SHA-256, byte size, MIME type, and dimensions when applicable. Shipped motion requires decoded, ordered PNG frame sequences with distinct initial, active, and resting states, exact CSS viewport/DPR metadata, at least 0.1% decoded-pixel change between adjacent frames, and a separately sampled reduced-motion stability sequence; a filename, URL, or unchecked media container is not proof. Font-file checks establish only the container signature. Computed-font reports and responsive specimens establish that the intended family and weights actually rendered. Remote URLs are references, not release evidence.

The manifest must validate against schemas/web-release-evidence.schema.json. It records three distinct direction artifacts, item-level scores, reviewed-copy hash, selected font and computed-font proof, required engineering reports, production commit, and rollback evidence.

Validate it:

npm run validate:release -- path/to/release-evidence.json --repo-root /path/to/owning-repo

Hard Stops

Do not build, merge, or promote when an applicable condition is true:

  • an existing URL has no current desktop and mobile capture;
  • recipient evidence is invented, inferred as fact, or copied from generic language;
  • a named partner surface lacks privacy and consent classification;
  • the first viewport hides the person, product, workflow, or offer;
  • typography has no role, specimen, source, license, or mobile proof;
  • motion has no named job or reduced-motion behavior;
  • proof consists only of claims, repository names, frameworks, or diagrams;
  • mobile is merely stacked desktop;
  • the same identity made, verified, and approved the release;
  • production cannot be inspected or rolled back.
  • the production receipt is embedded in the production commit it claims to prove.

Output

Return the experience thesis, selected direction, working preview, changed artifacts, editorial/visual/type/motion/accessibility/performance/privacy/ engineering evidence, independent verdict, production proof, and rollback.

Do not substitute a plan, score, CI result, or attractive screenshot for the complete release.

What ships with it

Read from the repository

Just SKILL.md. No reference files, no scripts.

Gives 0 of the 12 instructions most design frontend skills give in 938 tokens

Counted across 1,169 of the 1,878 authors here whose files we hold, read 2026-08-07

  • Use CSS variables for color consistencyin 72 of 1169, across 23 files
  • Commit to one bold aesthetic direction before codingin 72 of 1169, across 27 files
  • Match implementation complexity to the aesthetic visionin 70 of 1169, across 20 files
  • Add atmospheric background effects and texturesin 57 of 1169, across 9 files
  • Use unexpected spatial compositions and layoutsin 56 of 1169, across 8 files
  • Implement real working codein 55 of 1169, across 7 files
  • Vary themes and aesthetics across different designsin 48 of 1169, across 7 files
  • Launch chromium in headless modein 47 of 1169, across 4 files
  • Close the browser when donein 47 of 1169, across 4 files
  • Run provided scripts with help flag firstin 47 of 1169, across 4 files
  • Wait for network idle statein 47 of 1169, across 4 files
  • Use descriptive selectors for elementsin 47 of 1169, across 4 files

Said here and by no other author read

  • reconstruct context from brand and source material
  • capture existing mobile and desktop states
  • write an experience thesis
  • compare exactly three distinct directions
  • build the static composition before motion
  • pass editorial articulation checks

Grouped from the skills themselves: near-identical wordings counted once, and counted by distinct author, so one author publishing three of these counts once. Length counted with cl100k_base; the agent that loads this file may tokenize it differently.

Keep looking

Skills are one crate of 326,144. 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.