Ship validation
Skill Topurrra/claude-plugins/plugins/design-mastery/skills/ship-validation
Load before shipping any frontend output. The final gate: the distinctiveness mandate and the 15-item self-validation checklist. Run silently, rewrite until every answer is yes. If any answer is no, the work is not shipped.From its SKILL.md
npx -y skills add Topurrra/claude-plugins --skill ship-validationAssembled 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.
SKILL.md
5.6 KB, ~1.3k tokens by cl100k_base, as published. Nobody here has run it
ship-validation: the gate before shipping
Run this checklist silently before every output. Rewrite until every answer is yes. If any answer is no, the work is not shipped. This is the measurement instrument for the Principal Designer-Engineer identity.
Why this matters
The codex tells you what to avoid. The workflow tells you the process. This skill tells you when the work is done. "It looks good" is not a measurement. The 15 items below are binary: yes or no. You check, you rewrite, you check again. When all 15 are yes, the work ships. Not before.
The distinctiveness mandate
Before running the checklist, ask one question: "Could this interface belong to any other product?"
If the answer is yes, it is not done. Distinctiveness is not decoration; it is the product's character made visible. It comes from four sources:
- A color derived from the product's meaning, not the framework's palette. The accent color has a stated semantic reason: industry, heritage, emotional register, or contrast target. "It looks good" is not a reason.
- A typographic pairing chosen for character, not familiarity. The display face has a point of view. The body face is a workhorse. The pairing is deliberate, not "Inter for everything."
- A layout composition that reflects the content's actual structure. The layout is derived from what the content is, not from a template. A dashboard is not a marketing page; a settings page is not a hero.
- One or two signature details that recur across the product. A specific radius rhythm, a considered empty state, a distinctive data visualization, a signature interaction. These are the details that make the product recognizable as itself.
If all four are present and the interface still could belong to any other product, the distinctiveness is not strong enough. Push one detail further.
The 15-item self-validation checklist
Run this silently. Rewrite until every answer is yes.
| # | Question | Required answer |
|---|---|---|
| 1 | Is there zero purple-to-blue, indigo-to-violet, or cyan neon gradient? | Yes |
| 2 | Is the accent color chosen for a stated semantic reason? | Yes |
| 3 | Are all colors tokenized (no raw hex in components)? | Yes |
| 4 | Is there a declared type scale with more than one family or weight-role? | Yes |
| 5 | Is spacing drawn from a single scale, not ad-hoc? | Yes |
| 6 | Are radii varied by element role, not uniform? | Yes |
| 7 | Is the layout something other than centered symmetry, unless centering is genuinely correct? | Yes |
| 8 | Does the hero (if any) avoid the eyebrow-headline-subhead-two-CTAs template? | Yes |
| 9 | Are there zero em dashes used as stylistic tics? | Yes |
| 10 | Is the copy free of every banned phrase in the antipattern codex? | Yes |
| 11 | Is every interactive element a semantic element with keyboard support and a visible focus state? | Yes |
| 12 | Does body text meet at least 4.5:1 contrast (7:1 preferred)? | Yes |
| 13 | Are all animations under 320ms with a custom easing and a reduced-motion fallback? | Yes |
| 14 | Would this interface be recognizable as belonging to this product and no other? | Yes |
| 15 | Has every element been justified, or would removing it improve the design? | Yes |
If any answer is no, the work is not shipped. Rewrite the part that fails, then re-run the full checklist (a rewrite can introduce a new failure). The checklist is iterative until all 15 are yes simultaneously.
The rewrite protocol
- Identify which item(s) are "no."
- Rewrite the specific part of the output that fails.
- Re-run the full 15-item checklist (not just the fixed item).
- If all 15 are yes, ship. If any are no, return to step 2.
Do not ship with 14 of 15 passing. One failing item means the work is not done. The bar is 15 of 15, every time.
Common failure modes
| Failure | Fix |
|---|---|
| Shipping with 14 of 15 | The bar is 15. Rewrite the failing item. |
| Checking the fixed item only | Re-run all 15. A rewrite can introduce a new failure. |
| "Close enough" on contrast | Contrast is a number. 4.5:1 is the floor. Measure it. |
| "It is distinctive enough" | Ask the mandate question: could this belong to any other product? |
| Skipping the checklist because the task is small | Small tasks fail the checklist too. Run it every time. |
Red flags: stop and rewrite
- Any of the 15 items is "no." Stop. Rewrite. Re-run.
- You cannot state the semantic reason for the accent color. Stop. Find the reason or choose a different color.
- The interface could belong to any other product. Stop. Push one signature detail further.
- You are about to ship without running the checklist. Stop. Run it.
- You ran the checklist and are rationalizing a "no" as "close enough." Stop. It is not close enough. It is no.
Definition of done for this skill
- The distinctiveness mandate question is asked: could this belong to any other product?
- All four distinctiveness sources are present (color, type, layout, signature detail).
- All 15 checklist items are "yes," simultaneously.
- The full checklist was re-run after every rewrite.
- No item was rationalized as "close enough."
See also
design-mastery, the orchestrator that gates every output through this skill.antipattern-codex, the forbidden patterns this checklist enforces.designer-identity, the stance that demands distinctiveness over defaults.design-workflow, the step 7 where this checklist runs.
What ships with it
Read from the repository
Just SKILL.md. No reference files, no scripts.