Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
npx agentmods add skills/animalzinc/claude-plugins/design-referencenpx skills add animalzinc/claude-plugins --skill design-referencegit clone --depth 1 https://github.com/animalzinc/claude-pluginsWrote 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/animalzinc/claude-plugins/design-reference)<a href="https://agentmods.dev/skills/animalzinc/claude-plugins/design-reference"><img src="https://agentmods.dev/badge/skills/animalzinc/claude-plugins/design-reference.svg" alt="Measured on agentmods" height="20"></a>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 | $0.00000 | $0.01930 |
| Opus 5 | $0.00000 | $0.00965 |
| Sonnet 5 | $0.00000 | $0.00386 |
| Haiku 4.5 | $0.00000 | $0.00193 |
Grade A, and why
design-reference 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 — 235 lines — stays where its author put it; the contents beside it link to each section on GitHub.
name: design-reference description: "Iterative design research: brief → references → feedback → principles → prototype → approve. Use when starting a design phase for any frontend project, or when the user says things like 'design pass', 'design phase', 'collect references', 'design research'." disable-model-invocation: false allowed-tools: []
Design reference
Structured design research process that builds visual direction from reference review and stakeholder feedback. Produces a DESIGN_RESEARCH.md document and an approved prototype.
The problem this solves: Design conversations tend to be vague ("make it look modern") or skip straight to prototyping without establishing shared visual language. This process forces explicit reference review and feedback capture before any prototyping begins.
Arguments
$ARGUMENTS - Optional: path(s) to input files (voice memo transcripts, text notes, design brief). If not provided, ask user to describe their design direction.
Core rules
- Never skip ahead. No prototyping until principles are synthesized and approved (Step 5 complete).
- Capture everything. Record exact quotes and specific reactions per reference. "Like the brightness" and "like the layout" are different signals — both matter.
- Always checkpoint. Each step ends with explicit user confirmation before proceeding.
- Multiple rounds. Always do at least 2 rounds of references. Round 1 = broad exploration. Round 2+ = focused based on Round 1 patterns.
- Accumulate patterns. Maintain a running list of likes/dislikes/patterns that grows with each reference reviewed.
Display current step number prominently: **[Step X/8: Name]**
Process
Step 1: Receive brief
Read provided file(s) or ask user to describe their design direction. Extract:
- Project type: website, desktop app, mobile app, terminal UI, dashboard, etc.
- Vibe/tone: any aesthetic preferences mentioned (dark, bright, minimal, bold, etc.)
- Anti-references: anything explicitly disliked or to avoid
- Functional constraints: layout requirements, must-have elements, platform
- Existing design context: current visual state, colors in use, design system, related projects
- Who uses it: target audience
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 First seen · 235 lines · 0 tokens per session scan A 1ecb9389fe77
design-reference is a skill published in the GitHub repository animalzinc/claude-plugins (15 stars, last pushed 15d ago), licensed MIT. It costs nothing until one of its globs matches a file; then it loads 1,930 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
contentforge
Produce a publication-ready, fact-checked, brand-compliant, SEO-optimized content piece through the full 10-phase pipeline: every phase dispatched to a dedicated subagent (researcher, fact-checker, drafter, visual annotator, scientific validator, structurer, SEO/GEO optimizer, humanizer, reviewer, output manager)…
cf-style-guide
Import a brand voice profile from an existing style guide — a .docx/.pdf document, a URL, or manual input — extracting tone, formality, personality, approved/banned terminology, compliance guardrails, and author profiles into a structured brand-profile JSON at /.claude-marketing/{brand-slug}/Brand-Guidelines/, then…
cf-audit
Audit an existing content library for freshness decay and coverage gaps — scores every piece 0-100 on age, statistic currency, link health, and citation recency, maps covered topics against target keywords, and produces a prioritized report: top refresh candidates, gap topics, retire candidates, and the exact…
cf-brief
Generate a research-backed content brief from a keyword or topic — keyword data with volume and difficulty, top-5 competitor and E-E-A-T analysis, search-intent classification, audience pain points, a section-by-section outline with word counts and citation targets, plus SEO and AEO/GEO strategy (AI Overview status…
cf-help
Show the ContentForge user guide with live plugin state — version, agent/skill/command counts, connector availability, and runtime environment read from scripts/plugin-metadata.py, never hardcoded — plus pipeline overview, per-skill listings, worked examples, brand-setup methods, and a troubleshooting matrix. Triggers…
cf-publish
Execute CMS publishing: push a finished, reviewed piece (Phase 8 complete, quality score >=7.0) to Webflow or WordPress via MCP connectors as draft, scheduled, or live — always showing a full publish preview and waiting for your explicit yes/no/edit confirmation before anything is pushed. Runs the EU AI Act Article 50…