スキル cold-offer
📦
cold-offer
コンテンツリビジョン r1 高リスク 🌐 ネットワークアクセス⚙️ 外部コマンド📁 ファイルシステムへのアクセス🔑 環境変数⚡ スクリプトを含む
Draft Cold Offers Around Concrete Findings
Cold outreach often asks for a purchase before showing useful work. This skill structures a finding, a small solution sample, and a first email with local draft checks.
対応: Claude Codex Code(CC)
インストール前に停止して確認を求めてください。
計画を確認し、ファイルを変更する前にユーザーの明示的な同意を得てください。
自分のエージェントでインストール
このリクエストをエージェントにコピーしてください。正規の Skill ページとマニフェストが含まれています。
エージェントリクエスト
Review the Skillstore skill "cold-offer" from https://skillstore.io/skills/cmj-hub-cold-offer.md and its manifest at https://skillstore.io/api/skills/cmj-hub-cold-offer/manifest. Verify the artifact. Stop and obtain explicit user consent before installing or changing files.エージェントは引き続き計画を提示し、セキュリティポリシーで必要な確認を求める必要があります。
エージェントが読めるリソース
AI エージェント、クローラー、スクリプトがページ全体ではなく整理されたコンテキストを必要とする場合は、これらのリンクを使ってください。
テストする
「cold-offer」を使用しています。 Draft an offer about a homepage requesting meetings before showing any work.
期待される結果:
- Finding: Your homepage asks for a meeting before showing a work sample.
- Prototype: A one-page sample that presents the finding and a short trial first.
- First email: Your homepage asks for a meeting before showing any work. I prepared a one-page alternative. May I share the full report?
「cold-offer」を使用しています。 Check a complete draft whose first email says: Buy our retainer.
期待される結果:
- Result: Rejected because the first email contains a configured paid-product pitch phrase.
- Revision: Share the specific finding and the prepared sample instead of asking for a purchase.
「cold-offer」を使用しています。 Check a draft with a finding and first email but no prototype.
期待される結果:
- Result: Draft is incomplete.
- Revision: Add a nonempty prototype description before checking again.
セキュリティ監査
高リスクv1 • 2026/10/3 バージョン別レポートを開く
Most static findings are harmless documentation, CSS, encoding markers, or legitimate local tooling. Confirmed risks affect optional asset renderers: unsafe HTML insertion, inline script serialization, CSS interpolation, and unconditional frame directory deletion. No evidence found of prompt injection, credential theft, or malicious exfiltration; the Python scorer performs local validation without network access.
13
スキャンされたファイル
701
解析済み行数
3
レビュー項目
0
誤検知を無視
確認済みのセキュリティ上の懸念 (2)
高
Inline Script Breakout Through Specification Data
The demo renderer embeds JSON.stringify output directly in a script element. Supplied strings containing a closing script tag can terminate it and inject executable markup.
JSON.stringify does not escape HTML script terminators. The browser parses these serialized values inside an executable script element before DOM insertion.
高
Unchecked Color Value Can Escape Style Elements
Both renderers interpolate supplied accent_color values directly into style elements. A crafted value can close the style element and introduce scripts or external resources.
Neither renderer validates accent_color before HTML interpolation. Text escaping elsewhere does not protect this separate style-element context.
機能レビュー項目 (3)
これらは、このスキルに期待される可能性のある実際のローカル機能であるため、レビューが必要ですが、確認済みの悪意ある動作としてはカウントされません。
高
innerHTML assignment (XSS risk)
document.getElementById('lines').innerHTML = h;
Response lines from the supplied specification are concatenated into HTML without escaping. Malicious markup can execute browser event handlers during rendering.
高
innerHTML assignment (XSS risk)
document.getElementById('artwrap').innerHTML = (ART && t>=T_ART)
Artifact labels and lines from the specification are inserted as unescaped HTML. Crafted tags can execute browser scripts or initiate unwanted requests.
中
Synchronous file operations
import { readFileSync, writeFileSync, mkdirSync, rmSync } from 'node:fs';
The imported rmSync recursively deletes the output-derived frames directory at lines 102 and 119. Existing contents are removed without an ownership check.
リスク要因
🌐 ネットワークアクセス (6)
⚙️ 外部コマンド (18)
assets/card.mjs:4 assets/card.mjs:97 assets/card.mjs:21-65 assets/card.mjs:65-66 assets/card.mjs:66-87 assets/card.mjs:95 assets/card.mjs:99 assets/card.mjs:101 assets/card.mjs:107 assets/demo.mjs:5 assets/demo.mjs:116 assets/demo.mjs:15-97 SKILL.md:33 SKILL.md:41-44 SKILL.md:44-48 SKILL.md:48 SKILL.md:54-60 tests/test_score.py:14
📁 ファイルシステムへのアクセス (7)
🔑 環境変数 (2)
⚡ スクリプトを含む (2)
監査者: codex
このレポートを共有・引用
バージョン付き評価レポート、中立的なバッジ、埋め込みカード、引用を共有できます。Skillstore は証拠を報告しますが、この Skill が安全かどうかは判断しません。
レポートリンクをコピー
https://skillstore.io/skills/cmj-hub-cold-offer/audits/1?utm_source=security_passport&utm_medium=share&utm_campaign=versioned_reportMarkdownバッジ
[](https://skillstore.io/skills/cmj-hub-cold-offer?utm_source=security_passport_badge)HTMLバッジ
<a href="https://skillstore.io/skills/cmj-hub-cold-offer?utm_source=security_passport_badge"><img src="https://skillstore.io/badges/skills/cmj-hub-cold-offer/security.svg" alt="Skillstore security assessment" loading="lazy"></a>埋め込みカード
<iframe src="https://skillstore.io/embed/skills/cmj-hub-cold-offer.html" title="Skillstore Security Assessment" sandbox="allow-popups allow-popups-to-escape-sandbox" loading="lazy" referrerpolicy="no-referrer" width="420" height="180"></iframe>学術引用 (APA · BibTeX · CFF)
APA形式の引用
cmj-hub. (2026). cold-offer security audit report (audit version 1) [Author version unspecified]. Skillstore. https://skillstore.io/skills/cmj-hub-cold-offer/audits/1BibTeX形式の引用
@techreport{cmj-hub-cmj-hub-cold-offer-2026,
author = {cmj-hub},
title = {cold-offer security audit report (audit version 1)},
institution = {Skillstore},
year = {2026},
number = {1},
url = {https://skillstore.io/skills/cmj-hub-cold-offer/audits/1},
note = {Author version unspecified}
}CITATION.cff
cff-version: 1.2.0
message: "If you use this Skill, cite its author and this versioned security audit report."
title: "cold-offer security audit report (audit version 1)"
version: "unspecified"
type: report
authors:
- name: "cmj-hub"
date-released: "2026-10-03"
url: "https://skillstore.io/skills/cmj-hub-cold-offer/audits/1"
identifiers:
- type: other
value: "skillstore:cmj-hub-cold-offer:audit:1"
description: "Skillstore immutable audit report identifier"
Skillstore スコア
このスコアの理由 証拠の信頼度: 中 73
アーキテクチャ
85
保守性
87
コンテンツ
65
コミュニティ
83
仕様準拠
作成できるもの
Founder Outreach
Draft a first email that shares a specific public-page observation and a small sample instead of requesting a demo.
Consulting Offer Preparation
Package a verified business finding with a limited solution sample before discussing a larger paid engagement.
Team Draft Review
Check draft completeness and configured pitch phrases before a human reviews the message for sending.
これらのプロンプトを試す
Draft a First Offer
Draft a cold offer from this verified observation: [finding]. Include a finding, a small solution sample, and a first email without a product pitch.
Rewrite a Pitch
Rewrite this email: [draft]. Lead with [verified finding], name [prepared sample], and remove purchase requests, meeting requests, and demo invitations.
Validate a Draft
Review my draft at [local path]. Use the local scorer to check required fields and pitch phrases, then suggest revisions without sending anything.
Prepare and Review Variants
Create three offer variants from [verified finding] and [prepared sample]. Compare clarity, validate each locally, and explain limitations of the phrase checks.
ベストプラクティス
- Use verified public observations and describe only samples you have actually prepared.
- Review scorer results manually because fixed phrase checks can reject harmless text or miss sales pitches.
- Keep the core drafting workflow separate from optional asset renderers until their security issues are fixed.
回避
- Inventing recipient problems or claiming unfinished work is already available.
- Treating a successful scorer result as proof of accuracy, compliance, or effective outreach.
- Passing untrusted specifications to the optional renderers or reusing directories containing important files.
よくある質問
Which assistants can use this skill?
The marketplace lists Claude, Codex, and Claude Code. Running the local scorer requires Python 3.
Does the skill send the email?
No. It guides drafting and checks local input. You review and send the message through your own email tools.
What does the leak section mean?
It means a concrete business issue observed on a public page or product. It does not mean private or stolen information.
What does the scorer check?
It requires three nonempty text fields and rejects configured pitch phrases in the first email. It does not verify facts or assess writing quality.
Does this require paid APIs or network access?
The Python scorer uses the standard library and no network. Optional asset renderers use browser tools, external fonts, and additional dependencies.
Are the optional asset renderers safe for untrusted input?
No. They have HTML, script, and style injection risks. The GIF renderer also deletes an output-derived frame directory without checking ownership.