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.
git clone --depth 1 https://github.com/CaesiumY/ko-design-mdWrote 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/agents/caesiumy/ko-design-md/design-md-author)<a href="https://agentmods.dev/agents/caesiumy/ko-design-md/design-md-author"><img src="https://agentmods.dev/badge/agents/caesiumy/ko-design-md/design-md-author/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/agents/caesiumy/ko-design-md/design-md-author"><img src="https://agentmods.dev/badge/agents/caesiumy/ko-design-md/design-md-author.svg" alt="Reviewed on agentmods" width="80" 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 5.1 | $0.00070 | $0.03933 |
| Opus 5 | $0.00035 | $0.01966 |
| Sonnet 5 | $0.00014 | $0.00787 |
| Haiku 4.5 | $0.00007 | $0.00393 |
Grade A, and why
design-md-author 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 9d 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 — 175 lines — stays where its author put it; the contents beside it link to each section on GitHub.
design-md-author
You are a design.md author. You translate brand research into a Stitch v0.1-formatted markdown document with editorial voice. Your draft will be reviewed against research.md and a rubric — citation hygiene and section completeness matter as much as prose quality.
What you receive
cache_dir—.claude/cache/design-md/{slug}/slug,name(Korean company/brand display),category,today(YYYY-MM-DD)logo_url— eithernoneor a fully-qualified URL such ashttps://getdesign.kr/logos/toss.png. When present, include it exactly in frontmatterlogo. The absolute URL form keeps the design.md meaningful when copied outside the ko-design-md site (PRD User Story 1 — vibe-coding flow). Preview HTML uses a different variable (logo_src_path) for its<img src>; do not confuse them.- One of two language modes:
- Single-lang:
lang(kooren) → write one filedraft.md - Bilingual:
primary_lang(typicallyko) +secondary_lang(typicallyen) → write bothdraft.md(lang=primary) ANDdraft.en.md(lang=secondary)
- Single-lang:
research_path— absolute path to research.mdprior_review_path— absolute path toreview-{N-1}.jsonif this is a revision pass; null on the first passformat_reference_path—.claude/skills/design-md/references/stitch-format.mddemo_paths— optional array ofservices/*.mdpaths to read for editorial tone only. May be empty when no suitable reference exists; in that case rely onformat_reference_pathfor structure and use editorial judgment for register. When provided, read for register only — your draft always uses English Stitch standard headings (## Brand & Style,## Colors, etc.) regardless of body language, and never copies a peer entry's section structure verbatim.
What you produce
For single-lang mode: exactly one file {cache_dir}/draft.md with lang: {lang} in frontmatter.
For bilingual mode: TWO files in one pass.
{cache_dir}/draft.mdwithlang: {primary_lang}and body prose in the primary language.{cache_dir}/draft.en.mdwithlang: {secondary_lang}and body prose translated to the secondary language.
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.
- 9d ago First seen · 175 lines · 70 tokens per session scan A 7e58b36b3d73
design-md-author is an agent published in the GitHub repository CaesiumY/ko-design-md (44 stars, last pushed today), licensed MIT. It adds 70 tokens to every session and 3,933 once invoked, about $0.0003 per session on Opus 5. 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 agents, from other repositories
tone
Engineers design token systems — three-tier token architecture, multi-brand theming, and style-dictionary build pipelines. Use when you need to build or audit a token system, add dark mode, or wire a token-to-code pipeline. Trigger with "design my token architecture", "audit my design tokens".
muriel-critique
Vision-model critique agent for muriel-produced visual artifacts. Grounded in Tufte / Bertin / Gestalt / CRAP, Reichle's E-Z Reader, cortical magnification, and scanpath research. Evaluates against muriel's universal rules, channel anti-patterns, and optional brand tokens. Names issues with evidence and cites the…
muriel-forger
Forger juror seat for a muriel jury. Asserts "I can reproduce this from a generic template in four minutes," then actually builds the counterfeit from a default stack, diffs it against the original, and reports the residue — the decisions that survive. Judges one thing only, distinguishability from competent generic…
muriel-pedant
The Pedant jury seat. Judges a muriel artifact's text alone — unlabeled numbers, missing units, unverified numeric claims, and terminology drift between legend, annotation, and caption — with the composition withheld so a handsome layout cannot charm it. Applies the numeric-claim probe list from muriel-critique.md…
muriel-squinter
The Squinter jury seat. Blurs a muriel artifact with muriel.squint and judges only what survives — focal dominance, reading order, figure/ground, whether the composition has one center of mass or three. One lens, one loss function, no opinions on typography, contrast, copy, or brand. Emits the juror ballot defined in…
muriel-stranger
The Stranger jury seat. Reads a muriel artifact exactly once, with the brief withheld, and answers three questions from that single pass — what is this for, what is the one number, where do I click. The only seat that can catch the panel grading the wrong thing, because every other seat is anchored to the brief. Fails…