Based on RSS feeds curated by Andrej Karpathy, automatically fetchallentries, read the original articles, merge and sort by topic, and generate a structured Chinese daily report.
Language
Match user's language: Respond in the same language the user uses.
Data sources
RSS Pack address:https://youmind.com/rss/pack/andrej-karpathy-curated-rss
How It Works / Workflow
Execute the following steps in order. After completing each step, check the corresponding item, then proceed to the next step.
Progress:
- Step 1: Fetch RSS content
- Step 2: Create a material group
- Step 3: Full deep reading
- Step 4: Topic merging, sorting, and daily report generation
Step 1: Fetch RSS content
Use the WebFetch tool to fetch the RSS feed:
URL: https://youmind.com/rss/pack/andrej-karpathy-curated-rss
Parse the returned RSS XML/JSON and extract from the past 24 hoursallentries. Record each entry's:
- title
- link
- publication time (pubDate)
- source/feed name
- summary (description/summary)
Step 2: Create a material group
Create a material group file, naming format:
YYYY-MM-DD - Karpathy 精选 RSS 日报素材
Take theall RSS entries obtained in Step 1 and group them into this material group by source, without filtering.
Step 3: Full deep reading
Foreach RSS entry, use the WebFetch tool to read the original article link and obtain the full text or key paragraphs of the article. Do not filter; read all of them. Record:
- The article's core viewpoints
- Key data/findings
- Citation links
If the original article link is 404 or inaccessible, use the RSS summary as a substitute and mark it.
Multiple WebFetch requests can be initiated in parallel to improve efficiency.
Step 4: Topic merging, sorting, and daily report generation
Takeallentries and aggregate them by topic:
- Merge: multiple articles on the same or strongly related topics are merged into the same section
- Sort: arrange by relevance and importance, with core AI/technology topics first, and low-relevance topics (book reviews, historical retrospectives, non-technical content, etc.) later
- Generate: output the daily report document, naming format:
YYYY-MM-DD - Karpathy 精选 RSS 日报
Daily report template
Strictly follow the template structure below to generate the daily report content:
> Andrej Karpathy 精选的信源资讯汇总 | 共 [N] 条更新
---
## 🔥 核心主题：【主题标题】
【基于原文阅读的合并描述，包含关键观点、数据和发现】
- 来源：[文章标题1](原文链接1)
- 来源：[文章标题2](原文链接2)
## {emoji}【主题 2】：【副标题】
【合并描述】
- 来源：[文章标题](原文链接)
...（更多主题，按相关性降序排列）
---
## 📊 今日数据
- **XX** 条 RSS 更新
- **XX** 篇深度阅读
- **XX** 个主题：AA、BB、CC
## 💡 编者观察
【对今日整体趋势的简要评论，2-3 句话】
---
*本日报由 AI 自动生成 | 数据源：*[Andrej Karpathy curated RSS](https://youmind.com/rss/pack/andrej-karpathy-curated-rss)
Template rules
- Full coverage: every RSS update must appear in the daily report, with none omitted
- Multiple articles on the same topic are merged into the same section, with citation links listed separately
- Byrelevance sorting: AI/ML/technology core topics first, general technology topics in the middle, and non-technical content (book reviews, history, culture, etc.) later
- Start each topic with an appropriate emoji (🔥 for the most core topics, choose the rest yourself)
- The numbers in "Today's Data" must be consistent with the actual content
- All citations must include the original article link
- The daily report language is Chinese, with proper nouns kept in English
Complete the report
After the daily report is generated, output the following structured report:
[Karpathy RSS 日报] 完成！
日期：YYYY-MM-DD
RSS 条目：XX 条
深度阅读：XX 篇
合并主题：XX 个
输出文档：YYYY-MM-DD - Karpathy 精选 RSS 日报