Ai News Aggregator

AI/技术新闻聚合引擎。100+ RSS源并发抓取、兴趣评分、跨天去重、统一预取。

Security Scan

The SKILL.md claims local Python scripts and a JSON feed list but the published skill contains only SKILL.md (no scripts or install steps), so the package is incomplete and its runtime behavior depends on fetching external code or manual setup — proceed only after verifying the source and contents.

Purpose & Capability

The skill's description and commands assume a local package layout (scripts/rss_aggregator.py, rss_sources.json, arxiv_papers.py, etc.), but the registry entry contains only SKILL.md and no code files or install steps. Asking only for python3 is plausible for a pure-Python tool, but without the referenced scripts the skill as-published cannot perform its claimed functions. The homepage (GitHub) exists, which could explain missing code, but SKILL.md does not include explicit instructions to clone or install that repository — an inconsistency.

Instruction Scope

Instructions explicitly tell the agent (or operator) to run Python scripts at local relative paths and to edit scripts/rss_sources.json; none of those files are present in the package. The commands will perform network fetches from 100+ RSS sources (expected for an aggregator) but SKILL.md does not call out or constrain network access or unsafe content types. Because the archive is missing, it's unclear what the runtime scripts do (e.g., what HTML parsing, remote resource fetching, or external APIs they call).

Install Mechanism

No install spec is provided (instruction-only). That minimizes direct install risk from the registry, but also means the skill is non-functional as-published unless the user/agent acquires the code elsewhere (the homepage repo). If the code is fetched manually or at runtime (not described), that introduces higher risk depending on the source and method used to obtain it.

Credentials

The skill declares only a dependency on the python3 binary and requests no environment variables or credentials, which is proportionate to an RSS aggregator. One caveat: RSS sources list (rss_sources.json) can point to arbitrary URLs — the agent running the scripts will make broad outbound network requests, but no secret/environment access is requested by the skill itself.

Persistence & Privilege

The skill does not request always:true and does not declare any persistent system-wide privileges. It is user-invocable and allows autonomous model invocation by default, which is normal. There is no indication it modifies other skills or system config.

What to consider before installing

This package is incomplete as published: SKILL.md references local Python scripts and a rss_sources.json that are not included. Before installing or running anything, do the following: 1) Visit the provided GitHub homepage and confirm the scripts and feed list actually exist and match what's described. 2) Inspect the repository code (especially network calls, parsing, and any subprocess execution) to ensure it does only what you expect. 3) Verify whether additional Python dependencies are truly unnecessary. 4) If you must fetch and run the code, clone the repo locally and run it in a sandbox or isolated environment (no secrets, minimal privileges). 5) Review rss_sources.json to ensure it doesn't point to untrusted endpoints. If you cannot or will not review the upstream code, avoid running this skill because the registry package alone lacks the executable components it advertises.

AI News Aggregator — AI/技术新闻高性能聚合引擎

并发抓取 100+ RSS 源，12秒完成，支持 ETag/Last-Modified 缓存、日期过滤。

Setup

确保 Python 3.8+ 可用，无需额外依赖（纯标准库）。

When to Use

用户需要查看 AI/技术新闻、技术趋势、最新论文、GitHub 热门项目、AI 公司动态时使用。

触发关键词：

• "AI 新闻"、"技术新闻"、"今天有什么新闻"
• "最新论文"、"arXiv"、"AI 研究"
• "GitHub 热门"、"趋势项目"
• "OpenAI 动态"、"Anthropic 更新"

Architecture

ai-news-aggregator/
├── scripts/
│   ├── rss_aggregator.py      # 核心 RSS 抓取器
│   ├── rss_sources.json       # 100+ RSS 源配置
│   ├── arxiv_papers.py        # arXiv 论文搜索
│   ├── github_trending.py     # GitHub 热门项目
│   └── summarize_url.py       # 文章摘要
└── SKILL.md                   # 本文件

Data Sources

分类	源数	内容
company	16	OpenAI, Anthropic, Google, Meta, NVIDIA, Apple, Mistral 等官方博客
papers	6	arXiv AI/ML/NLP/CV, HuggingFace Daily Papers, BAIR
media	16	MIT Tech Review, TechCrunch, Wired, The Verge, VentureBeat 等
newsletter	15	Simon Willison, Lilian Weng, Andrew Ng, Karpathy 等专家
community	12	HN, GitHub Trending, Product Hunt, V2EX 等
cn_media	5	机器之心, 量子位, 36氪, 少数派, InfoQ
ai-agent	5	LangChain, LlamaIndex, Mem0, Ollama, vLLM 博客
twitter	10	Sam Altman, Karpathy, LeCun, Hassabis 等 AI 领袖

Core Commands

RSS 聚合

# 抓取所有源（最近3天新闻）
python3 skills/ai-news-aggregator/scripts/rss_aggregator.py --category all --days 3 --limit 10

# 只看公司博客
python3 skills/ai-news-aggregator/scripts/rss_aggregator.py --category company --days 1 --limit 5

# 只看中文媒体
python3 skills/ai-news-aggregator/scripts/rss_aggregator.py --category cn_media --days 3 --limit 10

# AI Agent 相关
python3 skills/ai-news-aggregator/scripts/rss_aggregator.py --category ai-agent --days 7 --limit 10

# 输出 JSON 格式
python3 skills/ai-news-aggregator/scripts/rss_aggregator.py --category all --days 1 --json

arXiv 论文

# 最新 AI 论文（按热度排序）
python3 skills/ai-news-aggregator/scripts/arxiv_papers.py --limit 5 --top 10

# 搜索特定主题
python3 skills/ai-news-aggregator/scripts/arxiv_papers.py --query "multi-agent" --top 5

GitHub Trending

# AI 相关热门项目（今日）
python3 skills/ai-news-aggregator/scripts/github_trending.py --ai-only

# 本周热门
python3 skills/ai-news-aggregator/scripts/github_trending.py --since weekly

Core Rules

1. 优先使用 --days 参数

默认抓取最近 N 天的新闻，避免获取过期内容：

• 日报：--days 1
• 周报：--days 7
• 月报：--days 30

2. 分类选择策略

用户需求	推荐分类
公司动态	--category company
技术论文	--category papers
中文资讯	--category cn_media
社区趋势	--category community
AI Agent	--category ai-agent

3. 缓存机制

• 首次抓取后自动缓存（ETag/Last-Modified）
• 缓存有效期 1 小时
• 重复抓取秒级完成

Configuration

编辑 scripts/rss_sources.json 添加/删除 RSS 源：

{
  "name": "OpenAI Blog",
  "url": "https://openai.com/blog/rss.xml",
  "category": "company"
}