Article Writer

文章写作助手 - 根据选题生成高质量文章

Security Scan

This is an instruction-only article-writing helper that is internally consistent with its purpose and requests no credentials or installs.

Purpose & Capability

Name and description ('文章写作助手') match the instructions: generate outlines and full article modules. The SKILL.md only asks for web_search/web_fetch for research, which is appropriate for an article-writing skill. No unrelated binaries, env vars, or config paths are requested.

Instruction Scope

Instructions direct the agent to run web_search and web_fetch to gather material and to produce highly sensationalized headlines, numeric claims, and '金句'. That is coherent for a marketing-style article writer, but the guidance lacks guardrails for attribution, source citation, plagiarism avoidance, and fact-checking; it also explicitly encourages strong, potentially unverifiable numeric claims and provocative language, which raises content-integrity and hallucination risk (not a security incoherence, but important to note).

Install Mechanism

No install spec and no code files — instruction-only. This minimizes disk/network install risk; nothing will be downloaded or written by the skill itself.

Credentials

The skill declares no environment variables, credentials, or config paths. It does reference web_search and web_fetch (agent tools) but does not request unrelated secrets or system access. The requested tooling is proportionate to the stated purpose.

Persistence & Privilege

always:false and user-invocable:true. disable-model-invocation:false means the agent could call the skill autonomously when eligible, which is the platform default; there are no other privilege requests or attempts to modify other skills or system configuration.

Assessment

This skill is coherent and low-risk from a platform/credential perspective, but review these practical considerations before installing: 1) The skill instructs the agent to use web_search/web_fetch — verify your agent's web tools and privacy settings, and be aware network calls will occur. 2) The SKILL.md encourages sensational headlines and numeric claims but gives no rules for citing sources or verifying facts; add guardrails (require citations, disallow fabricated numbers, enable fact-checking) to avoid hallucinations or copyright issues. 3) Because outputs are designed to be provocative and reusable, review generated content for plagiarism and legal/compliance concerns before publication. 4) If you prefer tighter control, limit this skill to explicit user invocation (don’t enable autonomous workflows) and test it in a non-production environment first. 5) No credentials or installs are requested, so there is minimal system/secret exposure from the skill itself.

文章写作助手

写作风格规范

开头模式

• 第一句：吸引眼球，制造反差或冲突
• 第二句：点明核心观点
• 示例： ✅ "我现在写一篇文章，从选题到全平台发布，只要10分钟。" ❌ "今天来讨论AI编程。"（太平淡）

标题公式

模式1：数字 + 成果 + 反差

• "10分钟完成全平台发布，效率提升94%"
• "从崩溃4671次到省50% token"

模式2：冲突 + 结果

• "Claude一个插件，让全球软件股蒸发2850亿美元"

模式3：疑问 + 答案

• "为什么我不再手动写文章？因为Agent比我先开始办公"

结构要求

• 模块化分段，用数字编号（01、02、03...）
• 每个模块500-800字
• 每500字必须有一句可截图金句
• 所有场景用 before vs after 对比

金句要求

• 短（10-20字）
• 狠（有冲击力）
• 可截图（视觉效果好）

示例： ✅ "把重复的交给系统，把判断留给自己。" ✅ "选题找我，不是我找选题。" ❌ "我觉得AI很有用。"（太平淡）

结尾要求

• "弹射"而不是"降落"
• 一句狠话收束
• 留下思考或行动指引

用词习惯

• 多用短句，少用长句
• 多用动词，少用形容词
• 多用数据，少用感觉
• 多用对比，少用描述

禁用词汇

❌ "非常"、"很"、"特别"（太虚） ❌ "可能"、"也许"、"大概"（不确定） ❌ "我觉得"、"我认为"（太主观）

工作流程

1. 收到选题
2. 搜索相关素材（使用 web_search 和 web_fetch）
3. 生成大纲（3-5个模块）
4. 按模块撰写
5. 添加金句
6. 优化标题
7. 输出Markdown格式

输出格式

# 标题

## 01 模块标题
内容...

## 02 模块标题
内容...

## 03 模块标题
内容...

---

总结金句