agentleFS
AkashPriyadarshiiAkashPriyadarshii/cookbook/design/SKILL.md

anti-default-gates

Skill0 starsChanged 3 months agoMITView on GitHub

Design/UX lesson: an anti-default ("don't use X color/pattern") gate must target the default reproduction AS A WHOLE, never a single element or hue family; and passing one gate must be necessary-not-sufficient so a generic alternative can't slip past. Use when writing anti-slop/sunset/hard-ban rules for a design skill or style guide.

The file

Files
SKILL.md

Anti-Default Gate Lessons

From a hard audit (design-genius) where 4 independent review lanes flagged a well-intentioned orange-ban as color-biased. Each one broke something real.

1. Ban the reproduction, never the hue

A rule "no orange accent" bans a color family outright — a heritage gold on warm ivory with a real derivation line (the gilt stamp on the product) is a legitimate design, and force-coloring it off is the bias. It also can't tell blanket harm from a justified use.

Fix: gate the default REPRODUCTION as a whole, all conditions AND-ed: fails only if (a) accent in the banned band AND (b) flat default bg AND (c) no causal derivation line. Clears if it escapes ANY one. Never ban a hue family.

Check: state the 3 conditions AND-ed; a correct use must clear it. If the gate says "never warm on cream," it's a hue ban, not a reproduction gate.

2. One gate passing ≠ the audit passed

A color gate only clears the COLOR axis. A site that passes ("generic blue, not in the banned band") can still fail type, layout, copy, motion — and a checklist that says "color-clear → mark DISTINCT, stop" lets the audit short-circuit before the other levers run. Silent false-negative.

Fix: make every gate's pass NECESSARY, NOT SUFFICIENT for the overall verdict. "Cleared the color gate" means only that horizontal check passed; run the full dimension rubric, and mark done only when every axis passes.

Check: does any gate's pass line end in "…and stop"? If yes, it can short-circuit the audit. Every gate must say what it clears and what remains.

3. Bless-with-variety, not one exemplar

A gate that names ONE passing example ("gold on ivory is DISTINCT") quietly blesses that example as the new safe default — the corpus re-converges on the crowned pass case instead of the banned one.

Fix: vary the exemplar across archetypes/hue families in the SAME clause (cool cobalt, green chlorophyll, warm gilt) so no single look is the blessed default, and make the derivation the thing that clears — not the hue.

Check: does the gate's pass-example repeat one fixed swatch every time? Diversify it.

4. Medium-aware tells

A web anti-slop checklist (centered hero, CTA line-wrap, hover, viewport) is nonsense on a physical/print/industrial surface with no cursor or scroll. Judging a print poster by "hover-only" is a false fail.

Fix: scope each tells-list to its medium. Web list → screen only; physical medium → its own form rules (material, shape, substrate). Never apply one surface's tells to another.

Check: would this tell even exist on a print poster or a device? If not, label it web-only.

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.