Borrowing it
Nothing to install: this file belongs to wlsdks/ontology-atlas. Take a copy, put it at the same path in your own repository, and replace the rules that are about this project with yours.
curl -O https://raw.githubusercontent.com/wlsdks/ontology-atlas/main/.agents/skills/design-system-audit/SKILL.mdgit clone --depth 1 https://github.com/wlsdks/ontology-atlasWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/skills/wlsdks/ontology-atlas/design-system-audit)<a href="https://agentmods.dev/skills/wlsdks/ontology-atlas/design-system-audit"><img src="https://agentmods.dev/badge/skills/wlsdks/ontology-atlas/design-system-audit/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/skills/wlsdks/ontology-atlas/design-system-audit"><img src="https://agentmods.dev/badge/skills/wlsdks/ontology-atlas/design-system-audit.svg" alt="Reviewed on agentmods" width="80" height="20"></a>- NVIDIA SkillSpector warn
SkillSpector: 3 findings, up to high
These are SkillSpector’s own severities. On a checked sample its high-severity flags on skills were ~96% false positives — a documented command, a public API, a “never do X” rule — so we show them as a caution to read, not a verdict. Why →
- high Anti-Refusal · line 15 Skill instructs the agent to omit warnings, disclaimers, or ethical commentary. Stripping safety caveats hides risk from the user and is a common jailbreak preamble.Fix: Remove instructions that suppress warnings, disclaimers, or ethical commentary. Let the agent surface safety-relevant caveats to the user.
- high Anti-Refusal · line 125 Skill instructs the agent to omit warnings, disclaimers, or ethical commentary. Stripping safety caveats hides risk from the user and is a common jailbreak preamble.Fix: Remove instructions that suppress warnings, disclaimers, or ethical commentary. Let the agent surface safety-relevant caveats to the user.
- medium Excessive Agency · line 124 Skill enables autonomous high-impact decisions without human-in-the-loop verification. Critical operations (destructive commands, financial transactions, data deletion) should require explicit user confirmation.Fix: Add human-in-the-loop confirmation for destructive, irreversible, or high-impact operations. Never auto-execute commands that modify files, send data, or alter system state.
What it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5.1 | $0.00000 | $0.01282 |
| Opus 5 | $0.00000 | $0.00641 |
| Sonnet 5 | $0.00000 | $0.00256 |
| Haiku 4.5 | $0.00000 | $0.00128 |
Grade A, and why
design-system-audit scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 5d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 128 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Audit the system, not one screen
A ramp is the allowed value ladder; an off-ramp value is hand-selected outside it; a gate is automated enforcement. A 2026-08-03 inventory found 300+ off-ramp values, but their cause was four gate holes:
| Hole | Effect |
|---|---|
| lint saw bracket values only | 268 named Tailwind steps such as text-sm and rounded-md bypassed the ramp |
| two central surfaces were warning-only with no warning cap | 66 violations blocked nothing |
| a colour checker skipped a directory | raw colours had never been checked there |
| one screen owned a parallel four-step ramp | 17 of 33 elements sat off the app-wide ramp |
Fixing values without closing entry holes guarantees recurrence. /design-audit
measures one finished DOM change; /responsive-sweep measures breakpoint bands;
this skill asks whether the system is enforceable.
Scope and requested action
Use the scope returned by pnpm design:route, or the user's explicit audit
scope. "Every" below refers to that scope, not every screen in the repository.
An audit request produces findings and repair proposals. Apply repairs only
when the task also authorizes them; the repair order below does not itself
authorize edits. Report unmeasured areas rather than expanding the audit silently.
0. Read actual ramps
Extract values from app/globals.css; do not audit from memory or a stale prose
copy.
grep -oE "\-\-text-[a-z-]+:\s*[0-9.]+px" app/globals.css | sort -u
grep -oE "\-\-radius-[a-z-]+:\s*[0-9.]+px" app/globals.css | sort -u
grep -oE "\-\-leading-[a-z-]+:" app/globals.css | sort -u
grep -oE "\-\-shadow-elevation-[a-z0-9-]+:" app/globals.css | sort -u
1. Measure gate reach
For every specification, ask what the detector cannot see.
- Syntax: bracket versus named utility, hex versus rgba, class versus inline style, prefixed variants, text arrow versus icon, direct value versus ternary.
- Path: lint the file directly and compare with an independent source search. Four app boundary pages once held 22 silent off-ramp lines outside scope.
- Level: a warning with no
--max-warningsis not a gate. - Exemption: inspect
shouldSkip*, allowlists, and ignores. Exempt exact files with reasons, never whole directories. - Parallel system: look for surface-local ramps differing from global steps by 1–4px.
- Letterform pin: reject gates that freeze an adjacent class or quote style rather than the property; they fail on legitimate rewrites.
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 5d ago Changed · +10 lines 6cb376b1fbf9
- 12d ago First seen · 118 lines · 0 tokens per session scan A e03bf2dda588
design-system-audit is a skill published in the GitHub repository wlsdks/ontology-atlas (103 stars, last pushed yesterday), licensed MIT. It costs nothing until one of its globs matches a file; then it loads 1,282 tokens. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.
Other skills, from other repositories
design-brief
Parse a structured design brief written in I-Lang protocol format into a concrete design spec. Eliminates ambiguity from vague requests like "make it professional" by requiring explicit dimensions: palette, typography, layout, mood, density, and constraints. Trigger keywords: "design brief", "create a design brief"…
fs-electric-studio
OpenDesign as an enterprise design platform: a buyer-forwardable proposal for a design-org's economic buyer — pain, value, ROI, rollout. Built as a decision-grade B2B sales deck for economic buyer, design VP, procurement.
critique
Run a 5-dimension expert design review on any HTML artifact in the project — Philosophy / Visual hierarchy / Detail / Functionality / Innovation, each scored 0–10. Outputs a single self-contained HTML report with a radar chart, evidence-backed scores, and three lists: Keep / Fix / Quick-wins. Use when the brief asks…
kami-landing
Produce a print-grade single-page kami (紙 / 纸) document — warm parchment canvas, ink-blue accent, serif at one weight, no italic, no cool grays. The output reads like a professional white paper or studio one-pager, not an app UI. Multilingual by design (EN · zh-CN · ja). One self-contained HTML file, zero dependencies.
open-design-landing-deck
OpenDesign's brand-story deck for partners and press: why it exists, what it believes, and where design is going. Built as a decision-grade marketing & GTM deck for partners, press, community.
fs-editorial-forest
Art-directing a fashion house's annual report — the editorial system, the photography rhythm, and the data spreads. Built as a decision-grade design craft deck for brand stakeholders, exec audience.