ai-news-crawler
Trigger phrase \"watching news!\"; Create a JSON file named with the current timestamp only in the current directory, crawl 10 AI-related news in descending order of time and output Chinese summaries
When & Why to Use This Skill
The ai-news-crawler is a high-efficiency Claude skill designed to automate the monitoring of the global AI landscape. It intelligently crawls leading tech publications and official AI blogs, extracts the top 10 most recent developments, and generates concise Chinese summaries. By delivering results in a structured JSON format, it eliminates manual searching and language barriers, providing a streamlined workflow for staying informed on artificial intelligence trends.
Use Cases
- Daily AI Briefing: Professionals can trigger the skill every morning to receive a curated list of the top 10 AI news stories, saving hours of manual browsing.
- Newsletter & Content Sourcing: Tech bloggers and journalists can use the structured JSON output as a reliable data source for creating AI-themed newsletters or social media content.
- Competitive Intelligence: Market researchers can track the latest announcements from major players like Anthropic and Google to monitor industry shifts and product launches.
- Cross-Language Knowledge Management: Users can bridge the gap between English-centric tech news and Chinese-speaking audiences through automated translation and summarization.
| name | ai-news-crawler |
|---|---|
| description | Trigger phrase \"watching news!\"; Create a JSON file named with the current timestamp only in the current directory, crawl 10 AI-related news in descending order of time and output Chinese summaries |
Purpose
Crawl the latest AI-related news from multiple websites, merge and deduplicate them, select 10 in descending order of time, unify summaries into Chinese, and write to a JSON file.
When to Trigger
When the request contains the phrase "Go, Little Plane".
Inputs
- optional keywords?: Topic keywords (optional)
- optional time_window?: Time window (default: 144h)
Output
- news_json_file_only
Constraints
- Create only one file: .json in the current working directory; Do not output any text or paths in the conversation.
- Each record must include fields: title, source, url, published_at (ISO8601), summary_zh, language (zh/en/mixed).
- Fixed quantity: 10 entries, sorted in descending order of published_at; Strict deduplication (using url and title as keys).
- Summaries for English content must be in Chinese; Retain the original language identifier in the language field.
- Avoid crawling paywalled or login-required pages; Exclude ads and sponsored content; Do not write sensitive information or keys.
- If the publication time is missing, use the explicit time on the page; If unavailable, exclude the entry or mark published_at: null and lower its priority—do not guess.
Candidate Sources
- The Verge / AI tag
- TechCrunch / AI tag
- Liangziwei (Quantum Bits) / AI section
- 36Kr / Artificial Intelligence tag
- Official Announcements / Anthropic, Google AI Blog (select latest announcements)
Steps
- Confirm time_window and keywords (if provided).
- Iterate through candidate sources, crawl recent content lists, and parse title/url/published_at and body summaries.
- Condense and translate English summaries into Chinese, maintaining neutrality and information density.
- Normalize timestamps to ISO8601, merge lists, and deduplicate by url/title.
- Sort in descending order of published_at, select the top 10 entries; Write to .json as a JSON array.
- Do not output any text in the conversation; Only create the file.
Output Example (Structure)
[ { "title": "示例标题", "source": "TechCrunch", "url": "https://example.com/news/123", "published_at": "2025-11-16T08:30:00Z", "summary_zh": "用中文概括要点，80-150字，覆盖背景、事件与影响。", "language": "en" } ]