用 Codex 或 Claude 帮你安装 复制这段 Prompt,粘贴到 Codex、Claude 或其他助手里,让它检查 Skill 页面并帮你完成安装。
直接命令不会经过审查 Prompt;运行前请先检查来源。
npx skills add https://github.com/nota-america/forgecat-agent-profiles --skill ce-promote命令会保持在同一行。复制前请横向滚动并检查完整内容。
想先保存到本地?可下载 SkillsMP 当前能够提供的文件。
AI(ChatGPT·Claude·Gemini 등)가 쓴 한글 텍스트를 "사람이 쓴 글처럼" 윤문해주는 오케스트레이터 스킬. 번역투·영어 인용 과다·기계적 병렬·관용구·피동태 남용·접속사 남발·리듬 균일성·이모지/불릿 과다 등 10대 카테고리 70개 AI 티 패턴을 탐지·분류해 내용은 한 글자도 건드리지 않고 문체·리듬·표현만 자연스러운 한국어로 재작성한다. shim의 route_hint(light|standard|heavy)로 경로를 정해 잘 쓴 글은 1콜, 표준은 2콜, 중증·장문만 3+콜(진단→겨냥 윤문→finalize)로 처리한다. 트리거 — "AI 티 없애줘", "AI 같은 글 자연스럽게", "GPT/ChatGPT 문체", "AI 번역투 고쳐", "사람이 쓴 것처럼 윤문", "AI 윤문", "ChatGPT 티 제거", "한글 AI 탐지·윤문", "AI 글 사람처럼", "번역투 제거", "영어 인용 많은 글 윤문", "AI 글 티 안 나게", "휴머나이저", "humanize Korean", "AI detector bypass 한글". 후속 작업 — "특정 카테고리만 다시", "윤문 강도 조정", "장르 바꿔서", "이 문단만", "2차 윤문" 도 모두 이 스킬. 단순 맞춤법·오탈자 교정은 직접 처리, 번역은 번역 스킬, 내용 추가·삭제를 동반한 재작성은 별도 집필 스킬.
가장 최근 윤문 결과를 2차로 다시 다듬는다 — 특정 카테고리·문단·강도 조정도 가능. humanize-korean strict 윤문(Phase B)을 기존 run_id에 재실행해 잔존 finding을 처리한다. 트리거 — "/humanize-redo".
AI가 쓴 한글 텍스트를 자연스럽게 윤문하는 진입 명령. humanize-korean 파이프라인을 Fast 모드(기본)로 실행하고 `--strict`면 정밀 3콜(진단→겨냥 윤문→finalize). 트리거 — "/humanize".
基于 SOC 职业分类
正在显示 SKILL.md
| name | ce-promote |
| description | Draft launch or promotion copy for a shipped feature. |
| disable-model-invocation | true |
| argument-hint | [optional: what shipped and/or channels, e.g. 'a tweet thread and a LinkedIn post'] |
Turn a feature that just shipped into copy-pasteable, user-facing announcement copy — right inside the engineering workflow.
After you ship, the messaging shouldn't wait for a separate marketing pass. ce-promote figures out what shipped, picks the right channels, and drafts the copy. It is spiral-agnostic by default: with nothing installed it draws on a lite layer of editorial and social-media expertise to produce strong channel-specific copy. When the Spiral CLI (see references/spiral-cli.md) is present and authed, it uses Spiral so the drafts are voice-matched to your brand — a subtle enhancement, never a requirement.
This skill drafts only. It never posts, publishes, commits, or opens PRs. Posting is a human action. The output is always drafts for you to review, edit, and ship yourself.
/ce-promote # Derive what shipped from context, draft defaults
/ce-promote [free-form description] # You describe what shipped
/ce-promote a tweet thread and a LinkedIn post # Request specific channels
/ce-promote 3 tweet options for the new export feature
If the user gave a free-form description of the feature, use it as the source of truth.
Otherwise, derive it from context (use what's available; don't block on any one source):
gh pr view --json title,body,url 2>/dev/null (and gh pr view for the current branch). The title and body usually state the user-facing value.git diff main...HEAD --stat and skim notable changes to ground the claim in what actually changed.[Unreleased] entry in docs/changelog.md, CHANGELOG.md, or similar.git log --oneline -15 for the arc of the change.Then write a 1–3 sentence summary of the user-facing value — what a user can now do that they couldn't before, and why they'd care. Describe the outcome, not the implementation. ("You can now export any report to CSV in one click" — not "Added a CsvSerializer and an export endpoint.")
If you can't confidently tell what shipped, ask the user one short question rather than guessing.
Default to a small, sensible set:
Scale to what the change warrants and to what the user asked for. If they named channels ("LinkedIn", "email", "a blog intro", "a short demo script"), draft those instead of or in addition to the defaults. A small fix needs one or two short drafts; a flagship feature can justify a cross-channel set. Don't force a fixed template.
First, detect Spiral's state with two quick, non-blocking commands:
which spiral
spiral auth status --json 2>/dev/null
Classify into one of three states:
which spiral finds nothing) → Path 0 (install), then Path A if set up, else Path B.spiral auth status --json:
"authenticated": true (equivalently "status": "authenticated") → Path A (voice-matched)."authenticated": false → Path 0, then Path A if the user signs in, else Path B.--json on auth status), fall back to the legacy signal in that same output: ready iff it contains spiral_sk_, else unauthed.Never let a Spiral failure, timeout, or odd output block or slow the skill — when in doubt, treat it as not-ready and continue.
When Spiral isn't ready, offer to set it up once — unless the user previously opted out. The point is one proactive nudge, never a recurring one, and never a blocker: a decline always proceeds to Path B. Any dismissal records the opt-out, so a single first-run decline stops the offer for good in this repo — the user is never asked twice.
Read references/spiral-cli.md for the exact setup prompt (built with the platform's blocking-question tool), the connect/install steps, and how the opt-out is recorded so later runs skip this. In short:
spiral login --json (CLI >= 1.8.0; non-blocking, the API key never passes through the agent). On status: already_authenticated → use Path A. On status: pending → surface the auth_url, the user approves in their browser, then poll spiral auth status --json until authenticated: true → Path A. Never have the user paste a key into chat. (Older CLI without agent login → suggest npm i -g @every-env/spiral-cli@latest, or have the user run spiral login themselves.) Escape hatch: "or the agent can just draft directly, without Spiral's personalization and humanization."Skip Path 0 entirely — straight to Path B — when the opt-out is already recorded, or when running headless / non-interactive (no human to answer). If a human is present but no blocking-question tool is available, do not skip — fall back to a numbered list of the two options in chat and wait for a reply (per the Ask section of references/spiral-cli.md).
Use the Spiral CLI so drafts match the user's brand voice. Read references/spiral-cli.md before composing the prompt — multi-channel vs. single-channel-variations is phrasing-driven (channel keywords / cue words vs. --num-drafts) and getting it wrong silently returns the wrong number or shape of drafts. The exact phrasing rules live there; don't restate them from memory. Essentials:
--instant and --json. Parse drafts[] (each carries its own channel) plus session_id.channel. Spiral decides how many drafts per channel — multi-channel runs often return several per channel — so never assume one-per-channel or drop extras.If the spiral write call errors or returns no usable drafts, silently fall back to Path B for the affected channels.
No Spiral needed — draft strong copy directly using a compact layer of editorial and social-media fundamentals. (The Spiral path goes further: brand-voice matching, humanization, saved styles, and cross-channel campaign orchestration.)
Editorial fundamentals — every channel:
Social fundamentals — distributed channels:
Per channel:
Drafts per channel: one strong draft by default; produce more only when asked ("3 tweet options"), capped ~3.
Show every draft as a clean, copy-pasteable block, labeled by channel. For each:
### X post
<the copy>
session_id and each draft's url so the user can open and tweak them in the Spiral web app.Single-channel variations — "3 tweet options":
User:
/ce-promote 3 tweet options for the new one-click CSV export→ Summarize the value. Spiral path:spiral write "3 tweet options for one-click CSV export" --instant --num-drafts 3 --json(no cue words). No-Spiral path: write 3 distinct tweets directly. Present all three.
Multi-channel set — "a campaign across X, LinkedIn, and email":
User:
/ce-promote draft a launch across X, LinkedIn, and email→ Spiral path:spiral write "announcing one-click CSV export — a launch across X, LinkedIn, and email" --instant --jsonreturns a set of drafts per channel (Spiral decides the count — often several), each carrying itschannel. (--num-draftsignored here.) No-Spiral path: draft one X post, one LinkedIn post, one email directly. Present every returned draft, grouped by channel.