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 skills add keenkim1202/saramchae --skill saramchae-kogit clone --depth 1 https://github.com/keenkim1202/saramchaeWrote 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/keenkim1202/saramchae/saramchae-ko)<a href="https://agentmods.dev/skills/keenkim1202/saramchae/saramchae-ko"><img src="https://agentmods.dev/badge/skills/keenkim1202/saramchae/saramchae-ko/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/keenkim1202/saramchae/saramchae-ko"><img src="https://agentmods.dev/badge/skills/keenkim1202/saramchae/saramchae-ko.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 | $0.00211 | $0.05965 |
| Opus 5.5 | $0.00084 | $0.02386 |
| Sonnet 5.5 | $0.00042 | $0.01193 |
| Haiku 4.5 | $0.00021 | $0.00596 |
Grade A, and why
saramchae-ko 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 2mo 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 — 447 lines — stays where its author put it; the contents beside it link to each section on GitHub.
saramchae-ko — Korean documents that read like a person wrote them
사람체: 사람 + 글씨체. The shape of writing a person actually produces.
A document fails in two ways at once. It sounds generated — 번역투, a 존댓말 sentence dropped into 평서체 prose. And it is shaped wrong — a 340-character paragraph, or nine nested bullets where two sentences belonged. The reader experiences both as the same problem, so this handles both.
For a non-Korean document use saramchae. The structural checks are the
same there; what is missing is everything in sections 4.1 through 4.3, which only exists in Korean.
Do not run both on one document.
1. Never touch these
Getting this wrong is worse than leaving the document alone.
| Zone | Why |
|---|---|
| Verbatim records — transcripts, logs, chat exports | Rewording destroys the thing they exist to prove. When a spec demands an "unedited" record, editing prose is editing. Only credential masking is exempt |
| Code fences, command output, stack traces | Line breaks are data there |
| Tables | The row is the unit. Never flatten a table into bullets to "simplify" |
| Numbers, IDs, hashes, paths, API field names, versions | Never paraphrase a value |
| Quotations from a source document | A requirement quoted from a spec stays word-for-word |
CLAUDE.md and agent instruction files |
Imperative voice and heavy bold are functional there. A 90% bullet ratio is correct for a file an agent parses |
Anything between <!-- saramchae:off --> and <!-- saramchae:on --> |
The author marked it. Do not edit it and do not report it — the scanner already blanks it |
Frontmatter is exempt from every check — a description field is one long line on purpose.
A document about bad prose is full of bad prose: a style guide, a changelog quoting a bug report,
a before/after specimen. That is what the saramchae:off marker is for. If you find yourself
reporting a document's own examples back to it, the region needs the marker rather than an edit.
What ships with it
1 file beside SKILL.md in the same directory: the scripts, references and assets a skill reads on demand. Not counted in the per-session cost; read them before you install if any of them is executable.
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.
- 2mo ago First seen · 447 lines · 211 tokens per session scan A 7f37bd037b50
saramchae-ko is a skill published in the GitHub repository keenkim1202/saramchae (5 stars, last pushed 2mo ago), licensed MIT. It adds 211 tokens to every session and 5,965 once invoked, about $0.0008 per session on Opus 5.5. 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 2026-08-31.
saramchae-ko compared with skills from other repositories
| Mod | Type | Stars | Last push | Licence | Scan | Tokens | Works with |
|---|---|---|---|---|---|---|---|
| saramchae-kokeenkim1202/saramchae | Skill | 5 | 2mo ago | MIT | A | 211 | Claude Code |
| speak-human-twRaymondhou0917/speak-human-tw | Skill | 1.1k | yesterday | MIT | A | 220 | Claude Code |
| aws-wechat-article-formattingaiworkskills/wechat-article-skills | Skill | 726 | 1mo ago | Apache-2.0 | A | 168 | Claude Code, Codex |
| qu-ai-weiLifelongLazyLearner/qu-ai-wei | Skill | 727 | 1mo ago | MIT | A | 119 | Codex |
| inkos-translationNarcooo/inkos | Skill | 11k | 13d ago | AGPL-3.0 | A | 35 | Claude Code, Codex |
| humanizer-rusmixs/humanizer-ru | Skill | 201 | 1mo ago | MIT | A | 263 | Claude Code, Codex |
| inkosNarcooo/inkos | Skill | 11k | 13d ago | AGPL-3.0 | B | 37 | Claude Code, Codex |