My

Generates daily tech news HTML reports from IT之家 and 快科技 with iOS liquid glass style, full text, images, and categorized sections.

Security Scan

The skill claims to fetch and assemble news from IT之家 and 快科技 but the included code only formats a pre-existing local JSON and local images into HTML — the claimed scraping capability is missing, and there are hard-coded local paths that don't match the described runtime.

Purpose & Capability

Description and SKILL.md state the skill will '抓取' (scrape) IT之家 + 快科技 and produce complete articles, but the shipped script (scripts/gen_html.py) does not perform any network requests or scraping. Instead it loads a local JSON and local images. Expectation mismatch: either scraping code is missing or SKILL.md is inaccurate.

Instruction Scope

SKILL.md instructs fetching RSS and list pages, downloading images, and constructing JSON, but runtime instruction to run python scripts/gen_html.py only builds HTML from an already-prepared JSON and image directory. The instructions ask the agent to access external sites if implemented, but the provided code will only read specific local file paths (hard-coded Windows paths). This is scope creep / inconsistency.

Install Mechanism

No install spec and only a single small Python script and a docs file are included. No remote downloads, package installs, or archive extraction are present — low install risk.

Credentials

The skill declares no required env vars or credentials (ok). However, the script hard-codes absolute Windows paths under C:\Users\choksta (DATA_FILE, IMG_DIR, OUT_FILE). That is unusual and may cause unexpected file reads/writes on a host where those paths exist; it also means the script won't work out-of-the-box elsewhere. No credentials are requested or used in code.

Persistence & Privilege

always is false and there is no install step that modifies agent/system configuration. The skill does not request persistent privileges or attempt to change other skills' settings.

What to consider before installing

This skill is inconsistent: it promises web scraping and automated collection but only includes an HTML generator that reads a local JSON and image directory. Before using it, confirm which behavior you expect: if you need automatic scraping, the scraping/downloading code is missing and must be provided or the SKILL.md should be corrected. Also edit DATA_FILE/IMG_DIR/OUT_FILE to sensible paths for your machine (the script currently points to C:\Users\choksta...). If you run the agent to perform the web fetches described in SKILL.md, be aware the agent would access external sites (IT之家 and 快科技) — verify you trust that network activity. Finally, since the script reads and writes local files, inspect or adjust paths to avoid unintended access to other user data.

tech-news-daily

科技资讯日报生成 Skill：从 IT之家 + 快科技 抓取内容，生成带 iOS 液态玻璃风格的 HTML 日报。

工作流程

1. 抓取数据（RSS + 文章页）

   • IT之家 RSS：https://www.ithome.com/rss/
   • 快科技列表页：https://news.mydrivers.com/（需按日期过滤 1113 开头链接）
   • 每条资讯获取：标题、正文、来源、图片 URL

2. 整理数据（JSON 格式）

   [
     ["key", "cat_id", "title", "src", "url", "text"],
     ...
   ]
   

   • cat_id: 1=手机/硬件, 2=AI/大模型, 3=汽车, 4=科技大厂, 5=系统App
   • text 需做 CJK+ASCII 空格处理：re.sub(r'([\u4e00-\u9fff])([A-Za-z0-9])', r'\1 \2', s)

3. 下载图片

   • 目录：news_imgs_today/
   • 文件名：{key}.jpg 或 {key}.png
   • base64 内嵌到 HTML

4. 生成 HTML

   python scripts/gen_html.py
   

   • 输出：tech_news_YYYYMMDD.html
   • iOS 液态玻璃风格：毛玻璃卡片 + 四色渐变背景
   • 正文完整展开，无截断
   • 图片按原始比例完整显示

数据格式

每条资讯为 6 字段数组：

• [0] key：唯一标识（英文，用于本地图片文件名）
• [1] cat_id：板块 ID（1-5）
• [2] title：标题
• [3] src：来源名称（如 "IT之家"、"快科技"）
• [4] url：文章链接
• [5] text：正文摘要（完整不截断，中英/数字之间有空格）

参考：references/data_format.md

设计规范

• iOS 26 液态玻璃风格
  • backdrop-filter: blur(24px) saturate(180%)
  • 背景 rgba(255,255,255,0.62) + 白色边框
  • 悬停上浮 spring 动画
  • 四色渐变背景（蓝→紫→绿→粉）
• 标题：完整显示，不截断
• 正文：完整展开，无 -webkit-line-clamp 限制
• 图片：width:100%; height:auto，原始比例完整展示
• 三栏自适应网格布局