A portable, tool-agnostic set of Agent Skills, prompt cards, and rules for using AI without producing generic, unverified slop.
Use it when you want AI to help you move faster while keeping humans responsible for taste, truth, context, and consequences.
AI slop happens when we outsource judgment. Good AI work happens when we use AI to accelerate drafts, options, and execution while humans stay responsible for the judgment. Use AI to move faster through the loop, not to skip the loop.
skills/ canonical Agent Skills (the product)
anti-slop/ core judgment guard
verify/ verify before claiming done
small-slices/ small, reviewable changes
deslop-code/ remove AI code slop (+ branch-diff mode)
deslop-writing/ make prose sound human (edit or detect)
deslop-design/ strip the generic AI look from UI
prompts/ copy-paste prompt cards + a live demo script
adapters/ cursor/ rules, claude/ install notes
The skills are the source of truth. Adapters are thin, tool-specific copies.
Paste the relevant skills/*/SKILL.md or a card from
prompts/anti-slop-prompts.md into context.
npx skills add the-hooman-loop/anti-slop-kit # all skills
npx skills add the-hooman-loop/anti-slop-kit --skill deslop-writingManual: cp -R skills/. ~/.claude/skills/. See
adapters/claude/README.md.
cp -R adapters/cursor/rules/. /path/to/your-project/.cursor/rules/See adapters/cursor/README.md.
| Skill | What it does |
|---|---|
anti-slop |
Core judgment guard against generic, unverified output. Keeps humans in the loop. |
verify |
Do not claim work is done until it is checked; report what was and wasn't verified. |
small-slices |
Small, reviewable changes that preserve existing design and naming. |
deslop-code |
Remove redundant comments, verbose names, one-use wrappers, dead defensive code. Optional branch-diff mode. |
deslop-writing |
Cut AI writing tells so prose sounds human. Edit (rewrite) or detect (flag) modes. |
deslop-design |
Subtract-first UI cleanup: one accent, hierarchy from scale and space, reject the default AI looks. |
See prompts/demo-script.md for a 5–10 minute
before/after walkthrough in any AI coding tool.
This kit ships first-party skills. If you want more depth in one medium, these public skills are worth a look (install counts as of writing):
- petergyang/no-ai-slop — writing detect/edit with a large named-pattern catalog (~7K installs).
- cursor/plugins
deslop— branch-diff code cleanup (~4K installs). - yetone/kill-ai-slop — subtract-first UI (~2K installs).
- samber/cc-skills
frontend-design-deslop— strategy-first UI, heavier question flow (~2K installs). - nexu-io/open-design
design-taste-frontend— landing/portfolio taste (~2K installs). - Anthropic
frontend-design— avoid the default AI aesthetics.
Adjacent, but critique/accessibility rather than deslop: ux-crit,
dataviz-crit, and a11y-audit review work; they don't rewrite it.
Ideas from the strongest of these informed this kit's skills. Skill files are not copied — they're linked here and credited.
MIT — see LICENSE.
