SuperAPI AI Directory
Partner with us中文AI newsFavorites
Skillsai-news-digest

ai-news-digest

General & other

View repository

Updated 10-16 08:35

About this skill

Call / Invoke / Call

Chinese

  • "Use ai-news-digest to generate today's AI news digest"
  • "Use ai-news-digest to get yesterday's AI news"
  • "Use ai-news-digest to see what AI developments there were the day before yesterday"
  • "Help me use ai-news-digest to organize recent AI news"

English

  • "Use ai-news-digest to generate today's AI news digest"
  • "Use ai-news-digest to get yesterday's AI news in English"
  • "Generate an AI news summary for 2026-01-20 using ai-news-digest"
  • "Use ai-news-digest to summarize recent AI developments"

Japanese

  • "Use ai-news-digest to create a summary of today's AI news in Japanese"
  • "Use ai-news-digest to get yesterday's AI news"
  • "Use ai-news-digest to summarize recent AI developments"
  • "Use ai-news-digest to tell me AI-related news from the day before yesterday"
Goal

Within the specified time window, obtain the latest content from a set of AI news sources and produce an "acceptable" news digest:

  • Deliverable: 1 Markdown digest (optionally also provide JSON data)
  • Each news item: title, time, source, link, 1-3 sentence summary, tags (optionally "why it matters")
  • Default strategy: RSS/Atom first, HTML as fallback; do not bypass paywalls
  • Default output:Chinese(if the source is in English, it must be translated into Chinese before output)
Input (ask clearly first)
  • Time window (default "today"):
    • Natural language:今天/昨天/前天(extensible 过去7天/最近24小时）
    • Specified date:YYYY-MM-DD(interpreted in the user's time zone as "that day 00:00-23:59:59")
    • Specific start and end time:since/until(including time zone)
  • Time zone: default UTC+8(recommended implementation default Asia/Shanghai), the user may specify explicitly
  • Output language: default zh(Chinese); optional en / both
  • Topic scope: research/product/open source/funding & investment/policy (multiple selections allowed)
  • Source scope: use the default list or a user-specified subset (see references/sources.md）
  • Output format: Markdown (default) / JSON / both
  • Article count limit: total and per-topic limit (default total 20)
  • Processing depth: title and summary only (default) / attempt body extraction (for non-paid content)
Process (recommended architecture)
  1. Load the source list (RSS first; for those without RSS, then do HTML adaptation)
  2. Fetch entries and standardize them (title/link/publish time/summary/source)
  3. Deduplicate and merge (same link/similar title/cross-source duplicates)
  4. Topic classification and ranking (time + source weight + keywords)
  5. Generate the digest (apply template assets/digest-template.md, attach source links)
  6. Quality check (time window hit, no duplicates, every item has a link/source)
Guardrails (compliance and stability)
  • Do not bypass paywalls/login restrictions; for paid content use only public titles/summaries/metadata
  • Respect site terms and robots; when necessary, degrade to "collect RSS entries only"
  • Failure can be degraded: output only the "obtainable" parts and list the failed sources and reasons
  • Translation/summarization must preserve traceability: always retain the original link and source in the output; when necessary, retain the original text in JSON title_raw/summary_raw
  • Before any file writing/overwriting export, first confirm the output path and whether to overwrite
How to run

CLI command

# 进入脚本目录
cd skills/public/ai-news-digest/scripts

# 获取今天的资讯摘要
python run.py --day 今天

# 获取昨天的资讯
python run.py --day yesterday

# 如遇本地 SSL 证书链问题（不推荐），可禁用校验
python run.py --day yesterday --insecure

# 指定日期
python run.py --day 2026-01-15

# 输出 JSON 格式
python run.py --day 今天 --format json

# 输出分享图片（需安装 Pillow）
python run.py --day 今天 --format image

# 输出横版图片（适合公众号）
python run.py --day 今天 --format image --image-preset landscape

# 输出浅色主题图片
python run.py --day 今天 --format image --image-theme light

# 写入文件
python run.py --day 今天 --out digest.md

# 使用 LLM 翻译（需配置 API key）
python run.py --day 今天 --llm

# 详细输出
python run.py --day 今天 --verbose

# 运行冒烟测试
python run.py --test

CLI Parameters

Parameter Description Default Value
--day, -d Date (today/yesterday/day before yesterday/YYYY-MM-DD) Today
--since Start time (ISO 8601) -
--until End time (ISO 8601) -
--tz Time zone Asia/Shanghai
--lang, -l Output language (zh/en) zh
--format, -f Output format (markdown/json/image) markdown
--image-preset Image size (portrait/landscape/square) portrait
--image-theme Image theme (dark/light) dark
--out, -o Output file path -
--topics, -t Topic filter (comma-separated) All
--sources, -s Source filter (ID, comma-separated) All
--max Maximum number of items 20
--max-per-topic Maximum number of items per topic 5
--llm Use LLM translation No
--verbose, -v Verbose output No
--insecure Disable SSL certificate verification (not recommended) No

Script Module Description

Script Function
run.py CLI entry point, integrates all modules
time_window.py Time window parsing
fetch_feeds.py Feed fetching (caching/rate limiting/retry)
parse_feeds.py RSS/Atom parsing and normalization
dedupe.py Deduplication and multi-source merging
classify_rank.py Topic classification and sorting
render_digest.py Markdown/JSON rendering
render_image.py Image rendering (social sharing cards)
summarize_llm.py LLM translation (optional)

Dependencies

Required(Python standard library):

  • Python 3.10+
  • Markdown/JSON output can run without third-party dependencies

Optional(enhanced features):

  • Pillow: Image rendering feature (--format image）
  • pyyaml: More complete YAML parsing (the script has a built-in simplified parser, and can load normally without installation sources.yaml）
  • anthropic or openai: LLM translation feature

Install optional dependencies

# 安装图片渲染支持
pip install Pillow

# 安装所有可选依赖
pip install Pillow pyyaml anthropic

Note：

  • When Pillow is not installed, Markdown and JSON output work normally, only image output is unavailable
  • When pyyaml is not installed, the script uses the built-in simplified YAML parser, and can normally load the complete source list
Resources
  • Data model and output specification:references/output-spec.md
  • Source registry:references/sources.yaml
  • Source evaluation description:references/sources.md
  • Time window specification:references/time-window.md
  • Topic keywords:references/topic-keywords.md
  • Translation specification:references/translation.md
  • Markdown template:assets/digest-template.md
  • LLM prompts:assets/summarize-prompt.md