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 jx-hxxx/hi-vibe --skill docs-keepergit clone --depth 1 https://github.com/jx-hxxx/hi-vibeWrote 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/jx-hxxx/hi-vibe/docs-keeper)<a href="https://agentmods.dev/skills/jx-hxxx/hi-vibe/docs-keeper"><img src="https://agentmods.dev/badge/skills/jx-hxxx/hi-vibe/docs-keeper/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/jx-hxxx/hi-vibe/docs-keeper"><img src="https://agentmods.dev/badge/skills/jx-hxxx/hi-vibe/docs-keeper.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 | $0.00231 | $0.04072 |
| Opus 5.5 | $0.00092 | $0.01629 |
| Sonnet 5.5 | $0.00046 | $0.00814 |
| Haiku 4.5 | $0.00023 | $0.00407 |
Grade B, and why
docs-keeper scanned grade B with 1 finding 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 1mo 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.
Strips warnings and disclaimersmediumAnti-refusal
Omitting safety caveats hides risk from the user and is a common jailbreak preamble.
internal init check; do NOT lecture the user about the marker or the How it starts
The opening of the file, as written. The whole thing — 258 lines — stays where its author put it; the contents beside it link to each section on GitHub.
docs-keeper
Each record has one job — never blur them:
| 문서 | 한 줄 정의 | 갱신 시점 |
|---|---|---|
CLAUDE.md |
코드만 봐선 모를 것 — 제약·함정·결정 (얇게!) | 그것들이 바뀔 때 (파일 이동은 해당 없음) |
<folder>/MODULE.md |
그 폴더의 상세 설계 | 폴더 내용이 바뀔 때 |
handover.md |
세션 맥락 (왜/어디까지 했나) | 세션 끝·compact 직전 |
CHANGELOG.md |
결과 (무엇이 바뀌었나) | 실질 변경 직후 |
evidence/METRICS.md |
증명 (정량 성과를 어떻게 재현하나) | 비교 가능한 실측이 생겼을 때만 |
Templates: templates/*.tpl in this skill's directory.
All user-facing prose you write into these docs follows the user's
conversation language — write in Korean if the user talks to you in
Korean, English if in English, and so on. Match the language they are
actually using with you; do not default to any single language. (An
existing doc's language wins: if the project's docs are already in one
language, keep writing in that one for consistency.)
Token budget (hard rule)
CLAUDE.md is loaded into EVERY session — every line costs tokens forever. Keep it ≤ 120 lines. Detail always goes DOWN into MODULE.md, never up. Never paste MODULE.md content, code, or long lists into CLAUDE.md.
Mode: init
- Detect state: does the project root have CLAUDE.md / handover.md /
CHANGELOG.md? Which top-level folders look like feature folders
(contain source files; skip
node_modules,.git,dist,venv,__pycache__, build output)? - If a file already exists, NEVER overwrite silently — show what is
missing versus the template and ask before touching it
(AskUserQuestion).
--auditmode: only report drift, change nothing. - Scaffold the LEAN default only — do NOT create every doc up front. A
toy project must not start with more management docs than code.
- CLAUDE.md: fill 개요/요구사항/실행 방법/함정/결정 기록 by asking the
user when not inferable. Write only what the code cannot tell you.
Do NOT generate a folder listing, a dependency list, or an
architecture overview — Claude can get those from
ls/grepin one second, they go stale immediately, and they cost context every session. (Claude Code's own/doctortrims exactly that kind of derivable content and keeps pitfalls, rationale and conventions — don't generate what the platform will delete.)상세 문서lists only MODULE.md files that actually exist — never write a pointer to a file you didn't create. It stays empty at init. - handover.md from its template.
CHANGELOG.mdfromtemplates/CHANGELOG.md.tpl(skip if it already exists — never overwrite). 트러블슈팅 기록이 이 플러그인의 핵심이라 CLAUDE.md·handover.md와 같은 급으로 처음부터 만든다. 지연 생성은 "첫 실질 변경 전까진 필요 없다"는 이유였지만, 그 결과 doctor는 없다고 경고하는데 init을 다시 쳐도 안 생기는 막다른 길이 됐다.- Do NOT create MODULE.md at init. 폴더별 상세 설계는 그 폴더가 실제로 복잡해졌을 때 생긴다 (step 3a). 처음부터 빈 MODULE.md를 뿌리면 아무도 안 채운 껍데기만 남는다.
- Do NOT create
evidence/METRICS.mdat init. 비교 가능한 전후 실측이 처음 생길 때만templates/METRICS.md.tpl로 만든다. 3a. Lazy docs — create each one the moment it is first needed, not before: <folder>/MODULE.md: create when that folder's design first needs recording — a structural change lands in it,reviewfinds it has grown complex, or the user asks "이 폴더 문서 만들어줘". Pre-fill 주요 파일 from a directory listing, leave 설계 as short TODO prompts, and add a line for it under CLAUDE.md's상세 문서in the SAME turn (per the doc-sync contract below). A file listing belongs HERE, next to the design it explains — not in CLAUDE.md, which every session pays for. (CHANGELOG.md는 더 이상 여기 없다 — init이 만든다. 이미 있는 프로젝트를 위해 log 모드는 여전히 없으면 만들지만, 정상 경로에서는 그럴 일이 없다.)
- CLAUDE.md: fill 개요/요구사항/실행 방법/함정/결정 기록 by asking the
user when not inferable. Write only what the code cannot tell you.
Do NOT generate a folder listing, a dependency list, or an
architecture overview — Claude can get those from
- Create the
.hi-vibe/marker directory — this is what turns the hooks ON for this project (the gate). Write.hi-vibe/initializedwith a one-line note (date + "hi-vibe initialized"), which also creates the dir. The gate is.hi-vibe/, NOT handover.md — a user may already have their ownhandover.md, so we must not key on it (that would misfire and could let a hook overwrite their file). Do not consider init complete until.hi-vibe/exists. - Ensure
.env*(see below),.hi-vibe/,.repo-xray/,handover.md,handover-archive.md, andhandover.md.lock(the transient write-lock file for concurrent sessions) are in.gitignore. The two dirs are caches;handover*is a personal session log kept local (not shared to GitHub). The other three docs — CLAUDE.md / MODULE.md / CHANGELOG.md — ARE committed. CLAUDE.md and CHANGELOG.md exist from init; MODULE.md arrives later, per step 3a.
What ships with it
5 files 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.
- 1mo ago Changed · +23 lines · +14 tokens per session 73cd49d403a0
- 1mo ago First seen · 235 lines · 217 tokens per session scan B 23b4857a42c5
docs-keeper is a skill published in the GitHub repository jx-hxxx/hi-vibe (3 stars, last pushed 1mo ago), licensed MIT. It adds 231 tokens to every session and 4,072 once invoked, about $0.0009 per session on Opus 5.5. A static security scan graded it B with 1 finding (strips warnings and disclaimers). No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.
Other skills, from other repositories
recipe-plan
Create work plan from design document and obtain plan approval.
atlas-present
Generate a polished HTML presentation page and Obsidian Canvas for big releases — new products, takeovers, major migrations. Non-technical audience. Use when asked to "present this", "release announcement", "show what we built", or "stakeholder update".
backlog
Manage the project's tech debt backlog. Add, list, fix, wontfix, delete, prioritize, and suggest batch actions on tracked issues, run by the user to manage accumulated debt (skills file their findings through /.zuvo/backlog-add themselves, never through this skill). Modes: list [category], add [description], fix B- …
retro
Engineering retrospective from git metrics. Reports deployment frequency, release cycle span, churn hotspots, backlog health. Outputs narrative report with 3+ actionable items. Flags: --since, --path, explicit range argument.
debt-ops-add
Register a deferred decision in the tech-debt registry. Trigger by judgment, not a marker scan, whenever a future reader would ask "why this way?": an unmade decision, stub, loosened type, bypassed check, swallowed error, a default picked "for now", or a TODO/FIXME/HACK/XXX marker. Trigger immediately whenever you…
ship-item
Ship a plan/todo item in a documentation-led repo — run the verify gate, integrate per the repo's model (fast-forward or PR), git mv todo→done with a shipped footer, advance the owning ADR(s) to Implemented, and regenerate INDEX. Use when the user says "ship this", "complete the plan item", "mark done", or "close out…